Closed Bug 123574 Opened 23 years ago Closed 19 years ago

Mail hangs while sending messages

Categories

(MailNews Core :: Backend, defect)

PowerPC
macOS
defect
Not set
critical

Tracking

(Not tracked)

RESOLVED WORKSFORME

People

(Reporter: brianclark44, Assigned: mscott)

References

Details

(Keywords: hang, qawanted)

Attachments

(7 files)

From Bugzilla Helper: User-Agent: Mozilla/5.0 (Macintosh; U; PPC Mac OS X; en-US; rv:0.9.8) Gecko/20020204 BuildID: 2002020411 On MacOS X, I will get at least four or five hangs per day, exactly like this: I compose a message, then I hit send. Then the "sending mail progress" dialog comes up. It first says "Sending mail". As soon as it switches to "Delivering mail", it hangs with the spinning cursor. It does not happen every time, but it happens far too often. I know the mail server that I am connecting to runs Sendmail 8.11.3. Reproducible: Sometimes Steps to Reproduce: 1.Run Mozilla for a while on MacOS X 2.Send and receive quite a few messages 3.Eventually, one will hang. I have a few reports generated from the "Sampler" application that comes with the MacOS X Developer tools, and I will post those reports here. Also, I will post more information on my website for this topic. http://home.attbi.com/~brian-clark2/index.html
Here is the first Report, generated with Sampler v. 1.1 on MacOS X 10.1.2.
Here is another hang that I had today. Same circumstances. Same version of Mozilla. I thought I would post two so one could be able to compare them.
Can you post an SMTP log of your hang? You can find directions for making a log on OSX here: http://www.mozilla.org/quality/mailnews/mail-troubleshoot.html#imap
Status: UNCONFIRMED → NEW
Ever confirmed: true
What machine and OS version is this? Are you running a dual CPU machine?
OK. This was taken with a build off of the 0.9.8+ branch. It should have the symbols included, and the Sampler report hopefully will be more helpful to you. Also, I am looking into the SMTP log thing. If I can get it to generate a log, I will post it here.
I am running MacOS X 10.1.2. It is a single CPU machine, a B&W G3 with a processor upgrade--a Sonnet G4 500Mhz. I also have 384MB RAM, if you need to know that.
Brian: thanks for the new stack. Interesting. Are you on some weird networking setup? PPoE over DSL, Linksys router...? Does this hang happen every time you send mail, or does it depend on the size of the mail message?
I tried creating an SMTP log, per the link that mscott posted. However, it doesn't seem to be working. I created a text file and dropped it on Mozilla to launch the app. However, after sending a few messages and quitting, I was not able to find the file I specified anywhere on my hard drive. Nor was I able to find any "seamonkey" folder. I even tried option-clicking on the Mozilla package and looking in there, but nothing that looks like a log. Is this supposed to work on MacOS X? Here is the entire text of the file that I dropped on Mozilla to start it up. Perhaps there is another way to startup Mozilla on MacOS X with these arguments? ARGS: -mail ENV:NSPR_LOG_MODULES=SMTP:5 ENV:NSPR_LOG_FILE=mozilla-smtp.log
See http://www.mozilla.org/build/mac.html#Environment You can use an 'ENVIRONMENT' file with those variables in.
No weird networking, at least not what I consider weird. I have a cable Internet connection from AT&T Broadband http://www.attbi.com I am connecting to this cable connection through a firewall from Sonicwall http://www.sonicwall.com and I do have NAT enabled on the Sonicwall, so my Mac has a private IP, which is translated to the public IP that my Sonicwall gets via DHCP from AT&T. It does not happen everytime I send mail. Only sometimes, and usually only after I have been using it for a while and sent a few messages. It doesn't seem to be related to the size of the message either, as it has happened to me when sending ordinary plain text email messages with no attachments.
What happens if you just wait for a long time? Does it eventually send, or just spin forever (5 mins or more)? Also, is there any console output during this time?
Keywords: hang
It never sends, no matter how long I wait. When it hangs, it really hangs. The entire application is unresponsive. I can't do anything but force quit. I will take a look at the console the next time it crashes. Also, I got the SMTP logging thing working, so I will upload that log as well after the next crash.
Interesting news to add to this today. I had a problem with my firewall and removed it for three days. During those three days, I did not crash once while in Mozilla Mail. When I put the firewall back in place, I crashed after using Mozilla Mail for only a couple of hours. My guess is that there is something in Mozilla Mail's networking code that doesn't like my firewall. It could be Mac specific, as I don't get this problem on Win2k when going through the same firewall. I also use Apple's Mail.app on the same computer, and it has _never_ crashed on me. I checked the console and it didn't yield much information. Here is what I got, with two entries, one before and one after the crash--nothing from the crash itself. One looks to be from Sampler, which I took from the crash. Feb 12 09:25:26 12-249-52-75 lookupd[209]: DNSAgent: dns_send_query_server - timeout for 63.146.85.72 2002-02-12 09:53:18.129 Sampler[509] *** Symbols couldn't be found for 40 addresses: 0x0050e280 0x00000001 0x0050e200 0x004fbe48 0x005026c4 0x005147ac 0x0050e108 0x0050ded8 0x0219cf0c 0x01bd0604 0x0050df50 0x005147b0 0x004ffd94 0x0050de84 0x0050e298 0x042b7bc8 0x0050e29c 0x004fbdb4 0x0050deec 0x02d6a120 0x004fbe60 0x0050dee0 0x0050dee4 0x005026bc 0x004fbe38 0x005147c8 0x0050df2c 0x004fbe24 0x0219bcbc 0x004fbd70 0x004fbe50 0x005026c8 0x01ec4df4 0x01c450fc 0x004ffda4 0x004ffdb8 0x0050de88 0x01d11dfc 0x004fbe14 0x00514dc8 I will post the Sampler report next.
Here is Console output that seemed to occur at the time of the crash. Hard to tell exactly, as the console didn't give the usual time/date stamp with these messages. *** malloc[517]: Deallocation of a pointer not malloced: 0xffffffff; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug *** malloc[532]: Deallocation of a pointer not malloced: 0xffffffff; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug *** malloc[508]: Deallocation of a pointer not malloced: 0xffffffff; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug *** malloc[508]: Deallocation of a pointer not malloced: 0xffffffff; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug *** malloc[508]: Deallocation of a pointer not malloced: 0xffffffff; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug *** malloc[508]: Deallocation of a pointer not malloced: 0xffffffff; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug *** malloc[508]: Deallocation of a pointer not malloced: 0xffffffff; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug *** malloc[508]: Deallocation of a pointer not malloced: 0xffffffff; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug DebugAssert: Third Party Client: chase->sleepQFlags == SLEEPQ_STATE_SLEEPOK [line 115, file Power/Power.c] *** malloc[508]: Deallocation of a pointer not malloced: 0xffffffff; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug 2002-02-12 12:09:18.613 Sampler[537] *** Symbols couldn't be found for 4 addresses: 0x00515670 0x00000001 0x0051567c 0x00515c64
sfraser just fixed some other crashes that would dump these messages to the console with mail (we were doing bad stuff when extra event loops were created, as with IMAP mail). try pulling today's bits and see if your problems are a little bit better.
This is probably some edge case in the Mac NSPR networking code that we're not handling correctly, that is tweaked by certain firewall settings. This is going to be hard to debug.
Yeah, let's hope you fixed it already :-) Brian, how long did you let it hang before performing a force quit? More than three minutes? Some network timeouts take that long, and in under special circumstances, it's possible they could lock the entire app for that period.
I normally don't wait 3 minutes, but I can the next time it hangs. I have tried waiting it out with earlier builds, but it just stays hung. Also, I will try using a build with sfraser's networking fixes on Friday, as that will be the first time I get a chance to get a new build. I don't have the ability to build one myself, so I usually just download the latest nightly when I need it. I do, however, have Apple's Dec. 2001 Dev tools installed on my machine. I can use any of them to help trouble shoot this bug. I don't really know much about them though. I would be happy to help in any way that I can. The networking fixes sfraser put in may or may not fix this, as I do use IMAP mail. These console messages could have been as a result of some IMAP operation, and not having to do with this crash.
With the latest hang I waited about an hour. It just stayed hung. Also, I can't get into the "Contents" folder of the nightly Mozilla package to retrieve the smtp log that I have been generating. It simply won't let me open the folder. Anybody know why?
Finder bug. Log out and log back in again.
Finally got one of these things generated.
This one is from build 2002021413. I assume that sfraser's fixes are in this build. I didn't see any console output at all after this hang, so perhaps that issue is taken care of. But it doesn't appear to have fixed my particular problem. I have hung three times today.
Cleaned up sampler data, showing our calls only: 586 nsSocketBOS::Write(char const *, unsigned int, unsigned int *) 585 PR_Write 584 pl_DefWrite 583 SocketWrite 583 SocketSend 581 MD_send 577 SendReceiveStream 347 OTLeaveNotifier [stuff deleted] 158 OTSnd [stuff deleted] 66 OTEnterNotifier [stuff deleted] 2 SendReceiveStream 2 SendReceiveStream [STACK TOP] 1 0x515664 1 0x515664 [STACK TOP] 1 0x51567c 1 0x51567c [STACK TOP] 1 ValidateProviderStruct 1 ValidateProviderStruct [STACK TOP] 1 pthread_mutex_unlock 1 pthread_mutex_unlock [STACK TOP] 4 MD_send 4 MD_send [STACK TOP] 1 SocketSend 1 SocketSend [STACK TOP] 1 SocketSend [STACK TOP] 1 pl_DefWrite [STACK TOP] 1 PR_Write 1 PR_Write [STACK TOP] I think we're stuck in the sending loop in SendReceiveStream. For some reason, we're not making progress.
I really haven't seen any mail hangs in quite a while. I don't know when it was fixed, exactly, but it doesn't seem to happen in Mozilla 1.0 or in Netscape 7 PR1 on MacOS X. Unless someone else is experiencing this, then I think this can be marked "resolved".
I have been having this problem and have spent a lot of time trying to figure out what circumstances cause it for me. I am very suspicious of certain originating mailers but have not found reproducable data. I can say: I NEVER have problems sending new mail that I have created. I frequently have problems replying to mail. Some recent builds seem to work better than others. I use every other nightly builds hoping for a breakthrough. Various measures will increase the likelyhood of the mail sending (I rarely use method #1 but follow the other steps in order): 1. Hit reply compose reply but delete the incoming mail message before sending - works everytime - conclusion ... there is something in the incoming HTML code that messes up the sending routine? 2. Under Option, Rewrap the entire message. Then send - sometimes works. 3. Copy the entire document into composer (command 4), copy all, paste back into "Commpose" (HTML code weirdness gets cleared up?) Sometimes works. 4. Save the message then send. Sometimes works. 5. Save the message, close, go into drafts, open message and send. High probability of working. 6. Quit Mozilla, re-launch Mozilla, go immediately to Drafts - likely to work. 7. Quit mozilla, launch Mozilla 1.1, go immediately to Drafts - always works. I have also been suspicous of my SMTP mail server and have learned through this activity report how to generate an SMTP log. I will try that and some of the other tricks you have taught me through these reports.
What I didn't indicate in my previous report is that I am working with a Mac 10.1.5 Mozilla 20021008 and am now waiting on the Beta release in a few days to begin my documentation again. This problem caused me to leave Netscape and I have found improvement with Mozilla but this bug still drives me nuts several times a day.
Using build 20021001808 Mac 10.1.5 I did some testing today to check my suspicion that the likelyhood of a reply hanging when sending increases with the number of ASCII characters in the message. The technique involves repetitive sending to your own mail address Steps: 1, Launch Mozilla, 2. Within your address book set your email address to "Prefers to receive messages formatted as 'unknown'" 3. compose a short message eg: "Test message" 4. send the message to yourself 5. read the incoming message 6. reply to the message with "Nth reply" (where N is the number of replies!) 7. repeat from 4. until the sending "hangs" This will hang on my system in about nine replies Now, quit Mozilla, launch Mozilla, repeat the above steps but in addition to the "Nth reply" line add a block of text (cut and paste from anywhere) to each reply On my system I hang in about two replies Repeat the above selecting either TEXT or HTML for message creation. Sending HTML mail hangs more quickly than text mail. My definition of "hanging" is that after approximately five minutes of waiting for the Delivering Mail progress bar the message "Sending of message failed. The message could not be sent because connecting to SMTP server smtp1.sympatico.ca failed The server ...." appears. I have done 15 of these tests and they are highly reproducible. When messages go through they go immediately. This may explain why my method #1 mentioned in comment 26, works. Is there anything I can test to aid in fixing this bug? I am particularly puzzled about why so few people are reporting HANG problems.
I have observed that bug 174808 may be related to, but is probably not a duplicate of, this bug.
I owe Mozilla an apology. I have finally solved my "sending mail problem" and it has nothing to do with Mozilla. It has everything to do with the fact that I am on an ADSL service that uses PPPoE. I am also sharing this PPPoE service through a router. This problem does not apply to cable service or dialup access. The problem is in the computer connected to the network. Service providers using PPPoE want to see a Maximum Transmission Unit (MTU) of 1492. Most computers are set to 1500 because that is what Ethernet typically expects. How do you change your computer's MTU? Some operating systems are beginning to adjust it as required. I suggest a web search on keywords MTU, 1492 and your OS . There are shareware tools to make the changes otherwise you must enter the potentially hairy world of "the Registry" or UNIX commands. Your search will reveal more about the history and circumstances of the "bug" than you probably want to know. If you are having problems sending mail or accessing certain websites, change your MTU to 1492. I am no longer frustrated sending mail. Life is grand. My apologies Mozilla. How does a note like this get added to FAQ or Troubleshooting? Many people on ADSL must be having this problem. I am going to post this to several reported bugs because I know the frustration of not being able to send mail. My apologies for multiple postings but this is a revolutionary fix.
I too have been able to confirm that running 1.2.1 of Mozilla on W2K SP2, as of yesterday I have been unable to send any message without Mozilla hanging. The task manager reports 100% processor utilization which about 30% of that in kernel space, yet the disk activity is nil. If I check the SMTP server, I can see that it has completed the TCP handshack to the server. At first, I presumed it might be related to the size of my sent folder, was 350MB, not 110MB and still no success...
Running Win98SE and Mozilla 1.2.1, I had two hangs on sending today, no successful sends.
Win2000 and Build 2003010708 and the mail hangs too, when sending; but its not reproduceable, don't even know, how I can help there.
Win2000 and Build 2003010708 and the mail hangs again. I just can see that CPU has 100% because of mozilla and memory usage is extremely changing. in the beginning, there are small changes with 1k/sec followed by changes of +/-20k
Blocks: 193931
Could it be that we are getting an error message from the SMTP server, and that is causing us to stop responding and hang?
Reporter, please generate another SMTP log, this time using Mozilla 1.3 beta or later.
Keywords: qawanted
I've experienced the same problem today with Mozilla 1.2.1 on Linux: TCP connection established, 100% CPU consumption and no reaction. This also happened with the "Sent later" setting. In addition, Mozilla hung with the same symptoms when I tried to open the "Outgoing Server (SMTP)" or the "Advanced Account Settings" panel. The problem happened for every mail I tried to send. It seems to have gone away after I terminated the SMTP server Mozilla was trying to connect to (before it tried to connect). Upon the next try Mozilla gave an error message that it could not connect. After having restarted the mail server Mozilla is now working fine again.
I'm running Mozilla in XP. When sending mail, Mozilla would hang, cpu usage would go to 100%, and Page File Usage would continuously increase. I deleted 'xul.mfl' (2,433Kb), and now it works again. The new xul.mfl file it created is 2,108Kb. I recall another instance where Mozilla would hang on the splash screen; deleting xul.mfl fixed that as well. What's up with Mozilla and xul.mfl files?
See bug 169777 for the xul.mfl problem. It is fixed, except for perhaps a few extremely rare cases, in 1.3 final.
Product: MailNews → Core
Brian: "this bug is no longer a problem for me... I no longer use Mozilla, but can confirm that the bug does not exist in Thunderbird 1.5." marking WFM (In reply to comment #37) > I've experienced the same problem today with Mozilla 1.2.1 on Linux: > TCP connection established, 100% CPU consumption and no > reaction. This also happened with the "Sent later" setting. 100% cpu would be a different issue - eg. bug 189738
Status: NEW → RESOLVED
Closed: 19 years ago
Resolution: --- → WORKSFORME
changed to thunderbird V1.5 and linux meanwhile. Suppose anyway, that so many things changed meanwhile, that this bug should be closed and a new one is opened, if this effects still are active.
Product: Core → MailNews Core
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: