Closed Bug 163951 Opened 22 years ago Closed 19 years ago

Stalls for 30 sec at copying sent mail to IMAP folder on sending

Categories

(MailNews Core :: Networking: IMAP, defect)

x86
All
defect
Not set
major

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: erich, Assigned: Bienvenu)

References

Details

(Keywords: fixed1.8, perf)

Attachments

(3 files, 1 obsolete file)

User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.1b) Gecko/20020821 Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.1b) Gecko/20020821 This bug was opened before and then closed. I just tested the current build and sending e-mails is very very slow. My server is Sun One Messaging server 5.2 Reproducible: Always Steps to Reproduce: 1.Send an e-mail (doesn't matter how big the e-mail is) 2. 3.
Sending emails is always fast for me. It could be network delays, server delays... Try using another email client, or better yet telnet to the smtp server and attempt to send something. If it's slow, then it's not mozilla.
erich@ebremer.com -- Have you tried the suggestions of bugs@zzxc.net? Unfortunately, your bug report doesn't contain enough information for us to be able to fix the bug. Could you please read the Bug Writing Guidelines at <http://www.mozilla.org/quality/bug-writing-guidelines.html> to see the kinds of information we need in a bug report. Please report back with more information after reading those guidelines? Thanks for using the guided form -- we just need more information. Feel free to re-open this bug if you can reproduce the problem on a recent build and have more details of the situation, or perhaps have narrowed it down to what might be causing the problem. -M
Status: UNCONFIRMED → RESOLVED
Closed: 22 years ago
Resolution: --- → INVALID
This problem occurs with Mozilla 1.1 but does not occur with Netscape Communicator 4.79. The problem seems to be with the "copies to sent" portion of the e-mail transmission. If the "Place copy to Sent" option is turned off, the e-mail will go out within 5 seconds. When the "Place copy to Sent" is turned on with the copy going to the *IMAP server* selected, the copy portion of the transmission can take in excess of 30 seconds (and for a small message). If a local Sent folder is selected, the sending goes fast (< 5 seconds). EVERYTHING WORKS FINE THOUGH WITH NETSCAPE 4.79 so I do not think the problem is with the IMAP server itself which is actually a Dual Pentium III Dell PowerEdge 2500 with very little load. The problem only occurs with Netscape 6 & 7 and Mozilla. Emptying the 530 messages in my sent folder on the imap server also does not help. Just playing around with it now, the copy to sent process will go fast when you send as plain text only, but goes slow when you say to send as plain text and HTML at the same time. Several of my people are having problems with this. HELP!
Status: RESOLVED → UNCONFIRMED
Resolution: INVALID → ---
Ok, I found some additional info. Sending goes slow when you REPLY to a message with HTML text and then when you send it as HTML and Plain Text. If you send as Plain text only, it works fine. Goofier, if you FORWARD the message (with attachement inline) and then send as HTML and Plain Text, it works fine. There seems to be some difference with REPLY and FORWARD. - Erich
QA Contact: olgam → stephend
This complaint (based on latest comments the problem is in copying to the IMAP server) seems to be the same as what is described in bug 169608 (or vice-versa).
Blocks: 193931
Keywords: perf
Summary: Sending of e-mails (regardless of size) is slow! → Sending of e-mails is slow! (copying sent mail to IMAP folder)
*** Bug 169608 has been marked as a duplicate of this bug. ***
related: bug 123063
Summary: Sending of e-mails is slow! (copying sent mail to IMAP folder) → Stalls for 30 sec at copying sent mail to IMAP folder on sending
*** Bug 166676 has been marked as a duplicate of this bug. ***
Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4a) Gecko/20030326 This has been a long-standing bug in Mozilla 1.x. It seems that any message copy from local to an IMAP folder (SSL or non-SSL) takes an extremely long time to negotiate. To make matters worse, there is little to no status information (i.e. "moved/copied message #4 of 20" used to be the default behavior in Netscape 4.x). There have been so many dups on this that I think the original bug report is starting to be diluted! Half of this bug relates to extreme slowness, the other half relates to lack of notification to the user as to what Mozilla is doing while it appears to be "idle". Bug 80877 Bug 156471 Bug 169608
Confirming on basis of duplicates. Should be addressed in 1.4 beta. Is this really a Windows-only bug? Maybe it's in our file handling or networking code?
Status: UNCONFIRMED → NEW
Ever confirmed: true
Flags: blocking1.4b?
Keywords: nsbeta1
Mail triage: nsbeta1-
Keywords: nsbeta1nsbeta1-
Flags: blocking1.4b? → blocking1.4b-
I have noticed stalling while trying to send mail in at least 1.2 and 1.3 of Mozilla. It seems that the mail is never actually sent, and it only occurs when replying to a previous email. If I create a new email and send it to the same recipient it goes just fine. According to the progress bar, the stall usually occurs at the end of the send, but I have seen it occur at begiining and middle of the send. Sometimes, killing Mozilla completely and restrating it clears the problem, but not always. The problem seems to occur randomly. I have not noticed a pattern as to return address, time of day, or phase of moon. --jon
After playing around for a while I am starting to see a pattern (at least for my specific flavor of this bug). It seems the problem is related to quoting or attaching the original text of a previously received email. When it happens, it doesn't make any difference whether I hit the reply button and include the original email text automatically, or create a new message and manually copy the text into the new email, or include it as an attachment. I've also noticed that sometimes if I start shaving off the copied text it sometimes will work. If I include none of the original text then it seems to always work (note: I have a very limited sampling for this problem). Based on the evidence I suspect that there is some screwy character in the original text or its formatting (perhaps it contains some bad html) that is causing Mozilla or SMTP to balk. In any case, this bug is awfully annoying. BTW: As with the original bug report, I am using Windows 2000, specifically with SP3. --jon
This will be my last update (promise). The light bulb finally went on. I remembered that this problem has happened to me before. A couple years ago it got so serious that I called up the ISP, thinking it was their problem. They insisted that I create a new outgoing server entry, even if the info was the same. I did it and poof, outgoing email stopped stalling. Well, after having tried everything else this time, I created a "new" outgoing server account. Poof, the problem seems fixed, again. Note that the information in the "new" account is exactly the same as the old one, and installing the latest Mozilla 1.4RC2 did not correct the problem. --jon
I'm going to take a shot in the dark and throw this over to Composition. Here's why: The bug only happens when replying to an email that contains HTML. Perhaps composition is doing some I/O that doesn't fly well with IMAP. If it's not Composition, feel free to shoot it over to Networking: IMAP. Also, I don't think that this is critical -- no data loss or crash, here. Downgrading to major, though normal might be more appropriate. -M
Severity: critical → major
Component: Mail Window Front End → Composition
Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5a) Gecko/20030619 Local-to-IMAP message copy performance has improved a bit in recent builds. However, the problem of no status ticker is still the same--when you highlight, say, 100 messages in your local "Sent" folder and drag them to your IMAP "Sent" folder, nothing happens for about 5 seconds. Then the barberpole starts moving at the bottom for about 30 seconds (which is when the message moves are actually taking place), then it disappears. Then about 5 seconds later all the moved messages finally disappear from the list. This is a horrible way to inform the user that something useful is actually happening. We need the old Netscape 4.x behavior of IMMEDIATE status on the bottom (i.e "Moving message 3 of 100...").
Re: comment 16 That would be a separate bug, asking for a dialog. -M
I'm still having the problem and I've noticed that it sometimes happens even on emails without HTML. The only somewhat reliable workaround I have for this problem is 1) create a new SMTP account with exactly the same info as the previous one 2) set the new account as the default 3) restart mozilla. Once the problem occurs on a given email I have never had it go through until I execute the workaround. In the meantime, other email replies may or may not go through. Also, if others are having this problem then it is a major one, not a minor one. I agree that data is not being lost, it is just sitting in my Drafts folder for a long time until I can shake the chicken bones and utter a chant. Because the problem is random and concerns something as fundamental as email, users may loose confidence in the app's overall reliability. --jon
The problem still occurs now and then, and can be pretty tenacious at times. I have even noticed it occuring on new email, not just replies. I am beginning to suspect that it has something to do with the address and perhaps the mood of the mail server (timing issue?). In the most recent example the email was quite short and plain text -- nothing fancy. I have used the email address many times before. I even tried creating the message again and copying and pasting the text. The send mail still stalls. I can send test emails to myself and others and it goes through just fine. BTW: I tried enabling logging but the log captures nothing when the problem occurs. This is really frustrating because it is so random. Can anyone suggest a good alternative mail app that is compatible with mozilla saved mail and address book?
As further evidence that there is possibly something related to the address, the message length, and the sever timing (along with the phase of the moon)... Regarding the short plain text email I created that refused to be sent, I tried sending a really short message to the very same email address and it went through. I tried sending the "bad" message and it balked. I then removed the first two sentences and a blank line, and the message went through just fine. Note: Although nothing shows up in the log when a message balks, the send/receive lights go back and forth a few times as if the system is trying to send the email, but can't.
Just wondering what's happening with this bug. It's still there in both 1.5 and 1.6b, and it's a very annoying bug. I lose my copy of about 30% of my sent emails every day, the emails go, but no copy for me :( Seems to me that there is just no retry or something similar, when it works (the good 70% of the time), it happens fast. If it doesn't happen (copying to IMAP sent folder) within 10 seconds, then it never seems to happen, and the timeout isn't for another 5 or 10 minutes (perhaps longer). Thus I think it's a retry issue. The IMAP server is remote (different ISP).
Copying the mail to the Sent Items folder doesn't seem to work at all (inc. v1.6), so I'm using the following workaround: 1. Edit -> Mail & News account settings -> Copies & Folders 2. Uncheck "Place a copy in". 3. Check "BCC these email addresses" and type your own address. For #3, I use sent@myusername.fastmail.fm which puts the mail directly in my Sent folder. You'd need a mail service that supports this configuration, otherwise the mail will end in your inbox. Prog.
I suspect I could use mozilla to move it to sent using filters as it comes back into my inbox (from the BCC idea in comment #22), as the other filters work fine with IMAP folders, but it's a pretty ugly workaround. By the way, the bug is marked as Windows 2000 and I'm using Linux, so it's not windows only.
As per comment #23, changing OS to ALL. If anyone encounters this bug on other platforms (e.g. Mac), please note. Prog.
OS: Windows 2000 → All
The real downside of the workaround (using BCC to save a copy via your inbox) is that you loose a copy of all the people that were BCC'ed! This is a pretty fatal flaw in the work around I think, I've got to choose between a 30% chance of loosing my copy of the email all together, or a 100% chance of loosing the list of all the people I BCC'ed it to. Not nice! Does anyone know if Thunderbird has the same problem?
Just some additional information. I mainly use Mozilla with IMAP remotely (not on a LAN). I just got back on the Lan and tested it to see if this problem occurs when I am local, and it doesn't! This doesn't help me, but it might help someone solve the bug. If anyone has checked this bug in Thunderbird, can they please let me know, because I can't find any Thunderbird Linux binaries (!?). This bug is very serious for me, if anyone knows of any other GUI IMAP client that's as good as Mozilla for Linux, could they let me know, because I'm losing too many emails due to this bug :(
OK, this bug is in Thunderbird too. I just tested 0.4 for a while, and still lost a couple of emails, although it didn't seem to lose as many as I'm losing in 1.6 of Mozilla. I fear this bug is not being monitored, I emailed the person it was assigned to a few days ago and have heard nothing back.
I think an ethereal dump of the session might help, here. sspitzer hasn't said anything on this bug -- let's see if anybody's awake over in Networking: IMAP. :-) -M
Assignee: sspitzer → bienvenu
Component: Composition → Networking: IMAP
QA Contact: stephend → grylchan
An imap protocol log of a failure to copy to the sent folder would be helpful, Craig. Here are the log instructions: http://www.mozilla.org/quality/mailnews/mail-troubleshoot.html#imap When connecting remotely, are you going through a firewall or vpn or something, that might be dropping the connection?
Hi David, Thanks for the input. I'm running behind a NAT'ing firewall (Netgear dedicated box) with no outbound restrictions, and the other end is using port forwarding, so it's pretty direct. Anyhow, here is the log, the gap is the long wait before the save gives up. I don't have any other IMAP troubles. I have junk auto moved to Junk, everything works fine. But a very significant number of sent email saves fail (30-50%). 98310[8b809a8]: 8b7ff70:mail.ucw.com.au:S-INBOX:ProcessCurrentURL: entering 98310[8b809a8]: 8b7ff70:mail.ucw.com.au:S-INBOX:ProcessCurrentURL:imap://craig@mail.ucw.com.au:143/addmsgflags%3EUID%3E/INBOX%3E80554%3E2: = currentUrl 16384[810a860]: 95540e8:mail.ucw.com.au:NA:SetupWithUrl: clearing IMAP_CONNECTION_IS_OPEN 98310[8b809a8]: 8b7ff70:mail.ucw.com.au:S-INBOX:SendData: 23 uid store 80554 +Flags (\Answered) 213000[9079060]: ImapThreadMainLoop entering [this=95540e8] 213000[9079060]: 95540e8:mail.ucw.com.au:NA:ProcessCurrentURL: entering 213000[9079060]: 95540e8:mail.ucw.com.au:NA:ProcessCurrentURL:imap://craig@mail.ucw.com.au:143/appendmsgfromfile%3E/Sent: = currentUrl 98310[8b809a8]: ReadNextLine [stream=8b80c38 nb=50 needmore=0] 98310[8b809a8]: 8b7ff70:mail.ucw.com.au:S-INBOX:CreateNewLineFromSocket: * 1445 FETCH (FLAGS (\Seen \Answered) UID 80554) 98310[8b809a8]: ReadNextLine [stream=8b80c38 nb=27 needmore=0] 98310[8b809a8]: 8b7ff70:mail.ucw.com.au:S-INBOX:CreateNewLineFromSocket: 23 OK UID STORE completed 98310[8b809a8]: 8b7ff70:mail.ucw.com.au:S-INBOX:SendData: 24 check 98310[8b809a8]: ReadNextLine [stream=8b80c38 nb=28 needmore=0] 98310[8b809a8]: 8b7ff70:mail.ucw.com.au:S-INBOX:CreateNewLineFromSocket: 24 OK Checkpoint completed 213000[9079060]: ReadNextLine [stream=93291b0 nb=0 needmore=1]
Craig, what version of what client generated that log? If you're not running the latest code, there's more logging in 1.7, and the latest nightly thunderbird builds that might be helpful
OK, I'm now running build 2004020507. Here is the new log. 98310[8e75ac0]: 8a9e060:mail.ucw.com.au:S-INBOX:ProcessCurrentURL: entering 98310[8e75ac0]: 8a9e060:mail.ucw.com.au:S-INBOX:ProcessCurrentURL:imap://craig@mail.ucw.com.au:143/addmsgflags%3EUID%3E/INBOX%3E80556%3E2: = currentUrl 16384[810a840]: 9bfa410:mail.ucw.com.au:NA:SetupWithUrl: clearing IMAP_CONNECTION_IS_OPEN 98310[8e75ac0]: 8a9e060:mail.ucw.com.au:S-INBOX:SendData: 19 uid store 80556 +Flags (\Answered) 294919[894ae28]: ImapThreadMainLoop entering [this=9bfa410] 294919[894ae28]: 9bfa410:mail.ucw.com.au:NA:ProcessCurrentURL: entering 294919[894ae28]: 9bfa410:mail.ucw.com.au:NA:ProcessCurrentURL:imap://craig@mail.ucw.com.au:143/appendmsgfromfile%3E/Sent: = currentUrl 98310[8e75ac0]: ReadNextLine [stream=8e75de0 nb=50 needmore=0] 98310[8e75ac0]: 8a9e060:mail.ucw.com.au:S-INBOX:CreateNewLineFromSocket: * 1447 FETCH (FLAGS (\Seen \Answered) UID 80556) 98310[8e75ac0]: ReadNextLine [stream=8e75de0 nb=27 needmore=0] 98310[8e75ac0]: 8a9e060:mail.ucw.com.au:S-INBOX:CreateNewLineFromSocket: 19 OK UID STORE completed <30+ second pause> <Message "connection to mail.ucw.com.au timed out"> 294919[894ae28]: ReadNextLine [stream=894b0b8 nb=0 needmore=1] 294919[894ae28]: 9bfa410:mail.ucw.com.au:NA:CreateNewLineFromSocket: clearing IMAP_CONNECTION_IS_OPEN - rv = 804b000e 294919[894ae28]: 9bfa410:mail.ucw.com.au:NA:TellThreadToDie: close socket connection 294919[894ae28]: 9bfa410:mail.ucw.com.au:NA:CreateNewLineFromSocket: (null) 294919[894ae28]: 9bfa410:mail.ucw.com.au:NA:ProcessCurrentURL: aborting queued urls 294919[894ae28]: ImapThreadMainLoop leaving [this=9bfa410]
OK, at first glance, it just looks like the connection to the imap server is taking a long time to establish. This is happening outside the imap code - I wonder if something somewhere is limiting the number of connections you can make to the server. You're right that we have no retry - you should get an error and get prompted to return to the compose window. Does that not happen?
Yes this does happen, but it's not much use, because the message has already been sent, so if I click send again, it will go twice. I suppose this could be a server problem (I'm using university of washington server, a fairly old one from RH 7.3).
do you ever get this error when you try to open an imap folder? It seems like it should be just as likely to get this error opening a folder (that you haven't previously opened) as it does getting it copying to the sent folder.
Occasionally it hangs opening another folder, not very often, but it does happen. This doesn't really cause much of a problem though, I just re-open the inbox, and then re-open the other folder. This bug with not saving Sent emails actually causes lost [copies of] emails, which is very nasty.
Can I get an idea if this bug seems solvable any time soon. It's ruining my email experience, and if it can't be solved soon I think I'm going to have to find another email client (ouch!) I'd donate some money towards this bug if it helps, is there any way to do that?
it's on my list of things to do. My idea is to add an option to try again, when the first fcc fails. I hope to get this into 1.7b (but not 1.7a) Yes, you can pledge money by adding a comment to the bug, and redeem your pledge by donating on www.mozilla.org. My personal opinion is that something is rather wrong somewhere along the way in your setup, and that all I can do in Mozilla is do the retry I described above, but it still going to be rather ugly when it spins for 30 seconds before failing (that's rather out of my control, how long it takes to fail). You might try turning on nsSocket logging and see if we get told anything interesting about why the connection is failing. on windows, it would be something like: set NSPR_LOG_MODULES=nsSocketTransport:5 set NSPR_LOG_FILE=c:\socketlog.txt
Hey David, I'll do a socket log now. The 30 second thing is a bit of a red herring I think, I've got one sending dialog up now thats been there for about 10 minutes, maybe more. Sometimes it fails quite fast (10-20 seconds), usually it's a minute or more. If you have any idea's on what could be wrong with my setup, let me know. I'm a programmer (Java), and understand tcp/ip networking pretty well. Also, I run Linux on both the client and the server, so I can run whatever tools you like. Perhaps a TCP proxy would help, to see all the traffic? I've got one that I use to debug http stuff. Also, I've run mtr (Matt's traceroute), and I'm gettting about 2% packet loss to all sites (my ISP's experiencing a growth spurt and hasn't upgraded capacity yet :( ). This doesn't seem anywhere near high enough to explain the 50% failure rate, but I spose it could be related? I'll pledge US$100 to this bug.
Craig, are these packets really lost, i.e., not received by the server? Or do they get resent? If they're truly lost, IMAP Append is going to spin until you cancel, because the way Append works is that we tell the server, here comes 1000 bytes, and then we send 1000 bytes, and then we wait until the server tells OK, got the data. If the server only gets 998 bytes, it will never tell us OK. But this pre-supposes that we've actually established a connection. From the log you sent me, we never established a connection. It might be that in some situations we do establish a connection and start sending data, and then run into a problem, and spin. In that case, cancelling should work, though again, you wouldn't be able to get your copy in the sent folder. An IMAP protocol log of the latter case would also be helpful. Re more logging, it sounds like you know more about the possibilities there. I think you can turn on server-side UW imap logging, but I don't know the details there...
Attached patch proposed fixSplinter Review
this patch makes it so we re-prompt when an fcc fails, and if the user says OK, we redo the fcc.
Attachment #143239 - Flags: superreview?(mscott)
Thanks David. Sorry I didn't get back to you, I couldn't get the UW IMAP server to do any logging. Seems like all the debugging is only compile time settings !?! Anyhow, I couldn't really get the source and recompile the server (too busy + old server). I'll check out your patch when it gets approved, I assume there will be a post here about it. Thanks again David. My client is upgrading their mail server, and I'm putting Cyrus IMAP server on it, so I'll let this bug know if that fixes the problem (and if it doesn't, debugging should be easier).
Attachment #143239 - Flags: superreview?(mscott) → superreview+
Craig, you can generate a client-side protocol log by following these instructions: http://www.mozilla.org/quality/mailnews/mail-troubleshoot.html#imap But I'll check in this fix, tonight, I hope, and you can see what happens.
I've had almost this exact problem for a long time (since pre Moz. 1.0). Here's what I see, figured you might be interested in more perspective: IMAP Server: University of Washington IMAP version 4.7. Mozilla mail clients: Thunderbird 0.6 (20040502) or Mozilla 1.7b Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.7b) Gecko/20040316 I can almost always reproduce this problem, just not very consistently. I.E., I might go a whole day of using TBird or Mozilla mail and not have the sent mail problem happen, then the next day it'll happen almost every time I send a message. The behavior I see is this: 1. Compose a message 2. Click Send 3. Send dialog comes up and if I have the problem, it will state one of the following: "Sending authenticate login information" or "Copying message to Sent folder" (and the status bar is up to 100%) 4. Either of these messages will stay up with the Compose window behind it for who knows how long -- I don't generally wait for them to time out. 5. I generally will eventually get this message if I leave the dialog window up (paraphrasing): "Message has been sent, but copy to sent folder faild, Retry?" 6. Instead of waiting, I wait up to 30 seconds, then click the Cancel button, and it returns me to the compose window. At that point, I close the compose window and answer "Don't Save" to the prompt. Here's where this gets interesting: 1. The message is ALWAYS saved to the Sent folder; I don't think I've ever seen it not saved except as noted in #3 below. 2. If I attempt to display the contents of my "Sent" folder at this point (i.e., select it with a single-click), I get the hourglass mouse pointer, and I can see the list of cached messages in my sent folder in the message-list part of the screen, and I can select messages that I've previously sent and view their contents. However, I do NOT see the message that I just sent where the Copy-to-Sent-Folder failed, or "stalls". The only way for me to see that message is for me to close TBird/Mozilla Mail and reopen. At that point, I can get into the Sent folder; no hourglass; can see the message I just sent. 3. One caveat to #1 is a scenario setup like this: a. Send a message; works just fine, everything as expected. b. Send another message and run into the problem -- the "Copying to Sent" dialog stalls on the screen. c. Now, cancel the Send dialog, close the Compose window, answer "Don't Save". d. Compose another message, click Send. e. Without a doubt, this message will stall on the Copying to Sent dialog -- every single time; never seen it behave differently. f. While this message does get sent to the intended recipient, it DOES NOT GET SAVED to the Sent folder. I.E., if a previous message has stalled on the copy to sent, and you haven't restarted TBird/Mozilla mail, the next and any subsequent messages you send not only will stall at the copy to sent dialog, but also will never make it into the sent mail folder. 4. One way that I can almost ALWAYS reproduce, although there have been times it works just fine, is if I attach a file to the message. I pretty much know I'm going to be restarting the mail client if I attach a message -- the messages get sent and copied to Sent, attachments and all, but I'll almost always have to cancel out of the Copying to Sent dialog, then close and reopen the mail client. Other tidbits: I've tested this on a multitude of Windows PCs. Brand new installations of XP Pro with all patches etc., brand new installs of either Mozilla or TBird. I've tried it on Windows 2000 with all service packs and patches. I've tried deleting every possible tidbit of either the Mozilla or TBird install (including my profile), reinstalling, recreating the profiles etc., and still run into the same problem. I clean up my Sent folder by archiving the messages fairly often (1x per month), but just to be sure, I've emptied my Sent folder to make sure it's not an issue with the size of the folder, or how many messages are in it etc. I generally use SSL, but it fails equally with either. I'm willing to do some kind of IMAP/SMTP logging if need be, just point me in the right direction. Thanks for listening.
I'm pretty sure this is the same bug as bug 123063. And probably also bug 196095. I noticed that someone recently implemented a change that allows a retry for the save to the IMAP sent folder. Originally this appeared not to help in my case - the retry always failed if the original save failed. I recently updated to Moz 1.7rc2 (on XP pro) and switched to the fastmail.fm smtp server. I'm pretty sure the switch to the new smtp server has fixed this problem for me - no matter where I send from, the faster smtp performance (or maybe the different use of sockets, as I switched to TLS) seems to mean that the "save to IMAP sent folder" activity succeeds. Which seems odd. The other thing I noticed is that on the odd occasion that my IMAP save fails, the retry now generally works. And finally, the IMAP retry seems screwy - if I try to save a message to draft and it fails, the retry dumps it in the sent folder! Cheers Alan
Flags: blocking-aviary1.0?
need to minus for lack of test case and patch. if we get closer to those before tbird 1.0 renominate.
Flags: blocking-aviary1.0? → blocking-aviary1.0+
chofmann -- You said minus but gave it a plus.
minus'ing for now pending the log file from Craig. Note the fix in this bug was already checked in and we think it fixed the problem for most folks. Need Craig's log to learn more about remaining issues.
Flags: blocking-aviary1.0+ → blocking-aviary1.0-
Product: MailNews → Core
(In reply to comment #48) > minus'ing for now pending the log file from Craig. That was from six months ago, and six months *after* the last comment here from Craig. Presumably the log is not forthcoming. (In reply to comment #45) > I'm pretty sure this is the same bug as bug 123063. A fix for bug 123063 was checked in this weekend; a fix for bug 287658 was checked in a month ago. Alan Hart, volesky@colostate.edu, Craig O'Shannessy: please try a current nightly build of Mozilla or TB, whichever you're using -- be sure the build is dated May 1st or later before bothering to download, as the nightlies on ftp.mozilla.org are not getting regular updates.
Very encouraged to see the recent flurry of fixes forthcoming. Would the following be an appropriate place to get the nightly build to try? ftp://ftp.mozilla.org/pub/mozilla.org/thunderbird/nightly/2005-05-03-05-trunk/ Thanks Alan
yes, that's the place. Bear in mind, if the problem is because of your imap server,the stall won't go away, but the cancel behaviour should be improved.
Sorry I didn't get back with a client log. I can confirm that it doesn't seem to be server dependant. I am running Mozilla 1.8a2 and am having the same problem with both Cyrus and uw-imapd (losing sent mail, after long hangs). The retry behaviour helps, and saves most of the timeouts, but I still get infinate hangs (e.g. > 10 hour), with no way to make the client retry. Interestingly this same behaviour has NEVER happened to me when connecting using a LAN, only ever using the net. I'll try to download a nightly in the next couple of days and let you know if there is any improvement. Thanks very much, Craig
OK, I tried a quick test from the nightly whose URL I mentioned above in message 50. Start Thunderbird - it seems to load all my folders and accounts correctly. Compose new message, giving it a subject line and an addressee only. Hit Save button (my drafts folder is on an IMAP server). The activity bar runs back and forth but the save fails after a few seconds. A dialog appears asking if I want to retry. I accept. The save succeeds this time but Thunderbird actually dumps the message in my Sent folder. I compose a new near-empty message, then send it immediately. Send succeeds, though smtp seems to take a good few seconds. Save of the sent messge to the correct IMAP folder proceeds according to plan. I leave Tbird running for the time it takes to type this report so far - 5-10 minutes, then compose and send a new short message. Successful, though again, smtp seems to take 20 seconds or so. Now I leave it running for perhaps half an hour while I load up a game. When I quit the game and compose a new short message, smtp is slow but successful, it correctly attempts to reconnect to my IMAP to save the sent message; this then spins for about a minute or so at the "sending login information" stage until this times out, at which point I get an alert. When I agree to a retry, the progress box now reads "Copy failed" throughout the retry. Again, it times out with an alert. I agree to retry again and this time it appears to write to the Sent folder successfully, and quickly. The sent folder, however, does not display the message. At this point I decide that the server and client are no longer communicating. I quit thunderbird and reload it. I switch to the sent folder, and there still seems to be some problem contacting the server, so the message just sent is still not there. After clicking between a couple of different accounts, I get an alert: Mail server mailserv.net is not an IMAP4 mail server. I dismiss the alert. Finally, clicking back to my sent folder generates the message "opening folder" in the status bar, which usually seems to indicate that the server is talking to my client again. And hey presto, the sent message is there. I would say that overall this is better than before but still not really usable in "IMAP sent folder" mode (BCC for sent messages works much better). Unfortunately it looks to me as though this bug still exists, though bug 123063 and its dupes may be cured. Thanks for all the effort thus far, David and others.
PS the infinite hang is apparently cured, though I don't think that's what this bug was originally reporting. The problem in bug 196095 appears to persist too. David, reading your earlier comments again, perhaps you expected this at this point. Server details: * OK [CAPABILITY IMAP4rev1 UIDPLUS CHILDREN NAMESPACE THREAD=ORDEREDSUBJECT THREAD=REFERENCES SORT QUOTA IDLE ACL ACL2=UNION STARTTLS] Courier-IMAP ready. Copyright 1998-2004 Double Precision, Inc. See COPYING for distribution information. Running Thunderbird under WinXP Home.
I have to agree with: [Additional Comments From craig@ucw.com.au 2005-05-03 07:32 PDT] Craig explains the symptoms exactly. I hadn't realized something until his comment about LAN verses the Internet. I'm a unix admin for a Postfix/Courier IMAP server. We are a company of 5 employees, that host mail for ourselves and about 160 email clients on a beefy Dell PE 1750 server (dual 1.2GHz, 2GB RAM). The only IMAP clients are the 5 employees. Of the 5 employees the following MUA's are used: 2 x Thunderbird 1 x Outlook 2000 2 x Outlook 2003 We were hosting the email server in our office on a T1 line (155 POP3 clients) and all 5 employees connected via 100Mb LAN, until three hurricanes whacked Florida, so we moved the server to the Verio complex. Since the move ONLY myself and the other Thunderbird MUA have this problem of copying the sent message to the sent folder. It is not 100% reproducible, but seems to happen 80% of the time. Sometimes I'll restart Thunderbird and it will work for a few sent messages, then back to the problem. I cannot pinpoint anything that seems to give me a concrete understanding about the true nature of the problem. I actually agree with the developers about the problem possibly being the IMAP server or network, but what concerns me is the fact that the other MUA's NEVER experience the problem? If the problem is the IMAP server or the network, shouldn't they have experienced this at least once in the past 10 months? I also wanted to mention that this could be related to specific settings on the IMAP server like connection time outs and other settings to protect DoS attacks that Thunderbird isn't expecting? I would like to offer some of the developers an account on my mail server if it will help troubleshoot the problem, I could also send the IMAP server logs to compare against the client logs.
Thought this might help from my previous comment: sshed to my IMAP server and issued this command: $ telnet localhost imap Trying 127.0.0.1... Connected to localhost. Escape character is '^]'. * OK [CAPABILITY IMAP4rev1 CHILDREN NAMESPACE THREAD=ORDEREDSUBJECT THREAD=REFERENCES SORT QUOTA AUTH=CRAM-MD5 AUTH=CRAM-SHA1 IDLE STARTTLS] Courier-IMAP ready. Copyright 1998-2003 Double Precision, Inc. See COPYING for distribution information. $ rpm -qi courier-imap Name : courier-imap Relocations: (not relocateable) Version : 2.2.1 Vendor: (none) Release : 1.8.0 Build Date: Fri 05 Mar 2004 03:59:59 PM EST Install Date: Fri 05 Mar 2004 04:05:17 PM EST Build Host: localhost Group : Applications/Mail Source RPM: courier-imap-2.2.1-1.8.0.src.rpm Size : 861000 License: GPL Signature : (none) Packager : %{PACKAGER} Summary : Courier-IMAP 2.2.1 IMAP server Description : Courier-IMAP is an IMAP server for Maildir mailboxes. This package contains the standalone version of the IMAP server that's included in the Courier mail server package. This package is a standalone version for use with other mail servers. Do not install this package if you intend to install the full Courier mail server. Install the Courier package instead.
I'm running the following nightly "Gecko/20050503" (1.8a2). And the problem SEEMS to be gone. I can't be sure, because the problem only shows up every now and then. I'm working from home tomorrow, so should be sending heaps of emails. This should thrash it a bit, but so far it looks good. Thanks! Craig
OK, I would say this this bug is fixed for me (e.g. bug 123063 and 287658 which I seemed to have are fixed). I haven't lost a mail, or got a hang since I updated to the nightly (on the 5/5/05, three days ago). Thanks to whoever fixed it!
(In reply to comment #49) > Alan Hart, volesky@colostate.edu, Craig O'Shannessy: please try a current > nightly build of Mozilla or TB, whichever you're using OK: Alan claimed this (or a similar symptom) was still a problem (comment 53 & 54); Craig claims it appears to have been cleared up (comment 58); <volesky> never responded. Tony Pecora chimed in and claimed that he sees/saw the symptom described by Craig, but neglected to identify which version of TB. Alan, Tony: still seeing it with *current* (e.g. 1.1a) builds of TB? xref bug 206408, where a similar problem is getting attention again.
Tested under XP on TB 1.1 alpha 2. Imap server is still the following: * OK [CAPABILITY IMAP4rev1 UIDPLUS CHILDREN NAMESPACE THREAD=ORDEREDSUBJECT THREAD=REFERENCES SORT QUOTA IDLE ACL ACL2=UNION STARTTLS] Courier-IMAP ready. Copyright 1998-2004 Double Precision, Inc. See COPYING for distribution information. One 5-min test did not immediately replicate the problem. The problem cannot always be immediately replicated however. Can you give me a couple of days to try it? Thanks Alan
The bug has now been reproduced under 1.1a2 as follows: 1. Compose new message. Enter the contents and addressee but don't send. 2. Go and brush teeth. Electric toothbrush recommended. 3. Return to PC and hit send. 4. "Copy failed" 5. Dialog appears. There was an error copying the message to the Sent folder. retry? 6. 10+ retries (hitting the retry button) did not work.
Alan, does your imap server limit you to four connections from a given ip address? Have you tried going into advanced imap server settings and setting the number of cached connections to 4?
Hi David, I believe reducing the number of connections from 5 to 4 did improve reliability greatly for me in earlier versions. I now use 4 cached connections as standard (including during this test) and the problem still occurs. I would be happy to make you a logfile later today, probably mid afternoon PDT. Are these instructions (below) still good for TB logging? http://www.mozilla.org/quality/mailnews/mail-troubleshoot.html#imap Cheers Alan PS if anyone can put the instructions for making an IMAP log up a bit more obviously on the mozilla.org website, I'd be grateful. Googling for Thunderbird IMAP log or mozilla IMAP log did not seem to be delivering the goods last night.
Alan, those instructions are still good. Re protocol logging, yeah, they should be easier to find. Ultimately, we'd really like to build logging into the product, so you could turn it on from the UI. TIA for generating a log. You can e-mail it directly to me, if you like.
I've sent David Bienvenu an IMAP logfile as suggested (rather than carefully censoring it and then posting here). Cheers Alan
in Alan's log, what was happening when trying to run the append to sent folder url was that all his connections had died - but he had reached his max connections limit, so we didn't try to spin up a new connection. The fix is simply to recalculate the number of alive connections before we decide if we can spin up a new connection or not.
Attachment #194511 - Flags: superreview?(mscott)
Comment on attachment 194511 [details] [diff] [review] fix for problem in Alan's log this seems like a good 1.8b4 candidate as we have had others report similar issues and the risk is very low.
Attachment #194511 - Flags: superreview?(mscott) → superreview+
Comment on attachment 194511 [details] [diff] [review] fix for problem in Alan's log yes, this is a clear candidate for the branch - I'd like to see if it fixes Alan's problem first on the trunk (I think it will). I'm sure it won't fix everyone's problems.
Attachment #194511 - Flags: approval1.8b4?
my recent patch has been checked into the trunk and will show up in tomorrow morning's nightly trunk build.
please resolve this and get a trunk verification so we can approve for the branch. thanks.
Turns out that Alan's problem was that he had four accounts to the same server, and it was dropping the fifth connection attempt. This patch restores the errror message we used to put up in this situation (I think either a necko change, or switching to blocking reads in IMAP changed the error code we were getting for this situation, so I added a check for the error we're now getting from necko). Next, I need to figure out why we were spinning in this case. But putting up an alert should help a lot.
Attachment #194680 - Flags: superreview?(mscott)
Comment on attachment 194680 [details] [diff] [review] put up error msg when courier server drops connection David, is this the right patch?
Attachment #194680 - Attachment is obsolete: true
Attachment #194680 - Flags: superreview?(mscott)
Attached patch correct fixSplinter Review
wrong tree again...
Attachment #194702 - Flags: superreview?(mscott)
second fix checked in, for courier problem...
Status: NEW → RESOLVED
Closed: 22 years ago19 years ago
Resolution: --- → FIXED
Attachment #194702 - Flags: superreview?(mscott) → superreview+
Comment on attachment 194702 [details] [diff] [review] correct fix this is a no brainer for the branch but we should still let it sit for a day or so on the trunk.
Attachment #194702 - Flags: approval1.8b5?
Flags: blocking1.4b- → blocking1.8b5+
I believe bug 196732 (not all junk labelled as junk is moved) is caused by the same courier connection limit. At least in my case. Alan
Attachment #194702 - Flags: approval1.8b5? → approval1.8b4+
Attachment #194511 - Flags: approval1.8b4? → approval1.8b4+
both patches are now on the 1.8 branch.
Keywords: fixed1.8
Product: Core → MailNews Core
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: