818198 - Loan :sfink a linux64 b-linux64-hp-0024

Status: RESOLVED FIXED

(Infrastructure & Operations Graveyard :: CIDuty, task)

Description

[Steve Fink [:sfink] [:s:]]

I really need to get back to getting the rooting analysis builds working. Perhaps it would help to have a slave to poke at that can set up a mock build environment. It hard fixing problems when the path is to make some buildbot-configs or buildbotcustom changes, get them reviewed and landed, then do a sample push, but I don't know how to set up my own development mock slave that's close enough to production to be useful.

And, in case whoever reads this knows: is there a way to temporarily point a loaner slave at my own dev buildmaster? (Or rather, since I can tunnel SSH all I want, is it ok if I do so and are there any useful tips I should be aware of?)

Thanks

Comment 1

[Rail Aliiev [:rail]]

Login information sent by email. bld-centos6-hp-005 is loaned.

Assign the bug back to nobody once you're done.

Comment 2

[Armen [:armenzg]]

Hi Steve, are you still in need of this machine?
Fine if you still do.

Comment 3

[Steve Fink [:sfink] [:s:]]

Yes, I still need to use it. I may need it for another couple of weeks.

This is for releng-related work, not dev-related stuff, btw. That's why I've hung onto to it for so long. I know that's generally not what these loaners are for, so let me know if you'd like to switch me to something else. I just need a Linux box that can act as a slave and see the build network (or whatever it's called.)

Comment 4

[Justin Wood (:Callek)]

Steve, just a one month check in - are you still using this host?

Still fine if you are.

Comment 5

[Steve Fink [:sfink] [:s:]]

(In reply to Justin Wood (:Callek) from comment #4)
> Steve, just a one month check in - are you still using this host?
> 
> Still fine if you are.

I'm bad, because I kinda haven't touched it for nearly a month, but it just so happens that I very much need a slave exactly like this for another purpose right now.

So yes, I wanna keep it.

Comment 6

[Chris AtLee [:catlee]]

Friendly poke - do you still need this machine?

Comment 7

[Steve Fink [:sfink] [:s:]]

I'm very actively using it. (Rooting analysis, sadly, is affected by the compiler version and perhaps some other mystery aspect of the environment, so this box has been very useful for replicating and tracking down intermittent issues.)

Let me know if my long-term hogging of this machine becomes an issue. I could always set up my own mock environment, which would *probably* be good enough (the host OS would be different.)

Comment 8

[Justin Wood (:Callek)]

(In reply to Justin Wood (:Callek) from comment #4)
> Steve, just another routine check in - are you still using this host?
> 
> Still perfectly fine if you are.

Comment 9

[Steve Fink [:sfink] [:s:]]

I will be again shortly, so I'd prefer to keep it.

Comment 10

[Armen [:armenzg]]

Has anything change since last time? I'm just doing some house cleaning. Fine if you still need the machine.

Comment 11

[Steve Fink [:sfink] [:s:]]

Oops, this needinfo got lost in the shuffle.

I still need it. I've been using it as the shared "wtf tbpl" machine for spidermonkey -- whenever someone on the JS team hits something that breaks and is only reproducible on a build slave, I fire up a master (usually a try build master, occasionally a mozilla-inbound build master) on my laptop and hook this slave up to it to reproduce the problem. I'm unclear at this point on what if anything I'm using from the build VPN. If I had a suitably configured machine, I could probably run it on the office network instead. Would you guys prefer that?

I'll needinfo? Callek instead of Armen, based on his bugzilla name info.

Comment 12

[Justin Wood (:Callek)]

Steve, we'll likely keep asking you -- but it sounds like you actually want something a bit more permanent than our loaner project usually allows for....

While this is all fine and dandy for now, I'm going to redirect the needinfo to coop based on that and based on your c#11 answers.

So far you've had this machine for 7 months (I say as a data point only, explicitly not as a problem)

Comment 13

[Chris Cooper [:coop] (he/him)]

I have no particular issue with a long-term loan like this, especially for linux64 where capacity is much less of an issue due to AWS.

Comment 14

[Steve Fink [:sfink] [:s:]]

Ok, this has been reimaged. I guess it needs some post-reimage steps now, and whatever is needed for me to regain access to it?

Thanks. (Assigning to coop since he's buildduty next week.)

Comment 15

[Chris Cooper [:coop] (he/him)]

Slave is ready. Credentials sent out of band.

Comment 16

[Chris Cooper [:coop] (he/him)]

Please close the bug when you're done and we'll recover the slave. Thanks.

Comment 17

[Steve Fink [:sfink] [:s:]]

Cool, thanks! Looks like I'm back in business.

Comment 18

[[e-mail] (:bhearsum)]

Are you still using this machine, Steve?

Comment 19

[Steve Fink [:sfink] [:s:]]

(In reply to Ben Hearsum [:bhearsum] from comment #18)
> Are you still using this machine, Steve?

yes

Comment 20

[[e-mail] (:bhearsum)]

(In reply to Steve Fink [:sfink] from comment #19)
> (In reply to Ben Hearsum [:bhearsum] from comment #18)
> > Are you still using this machine, Steve?
> 
> yes

Still? I haven't seen a logon since Feb 2nd.

Comment 21

[Steve Fink [:sfink] [:s:]]

(In reply to Ben Hearsum [:bhearsum] from comment #20)
> (In reply to Steve Fink [:sfink] from comment #19)
> > (In reply to Ben Hearsum [:bhearsum] from comment #18)
> > > Are you still using this machine, Steve?
> > 
> > yes
> 
> Still? I haven't seen a logon since Feb 2nd.

Yes, but (1) it's running a buildslave that I push to via my laptop master pretty frequently, and (2) I leave a terminal window open to my screen session there at all times, so the logins are really just showing when my VPN connection got restarted.

If this is tying up hardware that you'd like to free up for something else, then I can certainly move off of it to another slave. I keep very little state on that particular slave. Or at least, only on that slave -- I do have stuff there, but I can recreate almost all of it on a new slave with little difficulty. I did it recently when I had the slave reimaged. I think some RPM building stuff is probably the only thing of value there right now. (/me goes to copy it off.)

Comment 22

[Steve Fink [:sfink] [:s:]]

> Yes, but (1) it's running a buildslave that I push to via my laptop master
> pretty frequently, and (2) I leave a terminal window open to my screen
> session there at all times, so the logins are really just showing when my
> VPN connection got restarted.

To expand on this, I should explain why I push to this slave rather than just doing a try push. I use this box for testing mozharness changes, so I use SSH port forwarding to expose my local mozharness hg repo and have the slave pull from that. (It's pretty nice; it'll pull whatever I have pushed in my mq, so I don't even have to make "real" commits or hg push like I would if I were testing on Cedar or somewhere. Though it does take substantially longer to pull over my DSL uplink.)

I also tend to log on and look through the actual bits and log messages, sometimes from inside the mock environment, sometimes outside.

Comment 23

[Justin Wood (:Callek)]

:sfink, hows the state here?

Do you want us to reclaim this because its no longer used, do you want us to reclaim only to loan it back to you (essentially giving it most-recent machine code via puppet) or do you want to retain it in its current state?

[I'm away next week so buildduty may have to pickup where my Q ends]

I'll also note that as of _now_ this host host has been PING DOWN for 31 days and since it was loaned I didn't notice, which is an indicator to me that its likely unused

Comment 24

[Steve Fink [:sfink] [:s:]]

(In reply to Justin Wood (:Callek) (Back from PTO Apr 28) from comment #23)
> :sfink, hows the state here?

I disappeared for a while (work week, spring break) and was also tied up with repairing some security vulnerabilities I introduced, so it hasn't seen much use. :-) But I'm using it today for bug 964169.

> Do you want us to reclaim this because its no longer used, do you want us to
> reclaim only to loan it back to you (essentially giving it most-recent
> machine code via puppet) or do you want to retain it in its current state?

Its current state seems fine to me, though it has been rebooted so it wiped my buildbot.tac again. (No big deal, I just put it back.)

> I'll also note that as of _now_ this host host has been PING DOWN for 31
> days and since it was loaned I didn't notice, which is an indicator to me
> that its likely unused

What's that mean? It always seems to be there when I attempt to shell into it. Though I can definitely believe there are 31 day gaps when I don't log in.

Comment 25

[Chris Cooper [:coop] (he/him)]

This slave just got renamed as part of the colo move in bug 1013035. New fqdn is:

b-linux64-hp-0024.build.releng.scl3.mozilla.com

May need accounts re-created.

Comment 26

[Steve Fink [:sfink] [:s:]]

Yeah, looks like it lost my ssh key, and the original password I wrote down doesn't seem to be working either. Can I get a password? Thanks.

Comment 27

[Chris Cooper [:coop] (he/him)]

(In reply to Steve Fink [:sfink] from comment #26)
> Yeah, looks like it lost my ssh key, and the original password I wrote down
> doesn't seem to be working either. Can I get a password? Thanks.

Email sent with credentials.

Comment 28

[Steve Fink [:sfink] [:s:]]

I got back in, thanks.

Comment 29

[Justin Wood (:Callek)]

steve, still using this?

Comment 30

[Steve Fink [:sfink] [:s:]]

(In reply to Justin Wood (:Callek) from comment #29)
> steve, still using this?

yes

Comment 31

[Chris Cooper [:coop] (he/him)]

Our quarterly check-in...

Steve: are you still using this?

Comment 32

[Steve Fink [:sfink] [:s:]]

(In reply to Chris Cooper [:coop] from comment #31)
> Our quarterly check-in...
> 
> Steve: are you still using this?

Uh, yes, sorry. Using it now for testing moving spidermonkey builds into the tree.

Comment 33

[Chris Cooper [:coop] (he/him)]

It's 2015, so I'd like to check-in to see if you still need your loaned slave.

If yes, great, just let buildduty know when you are done.

If you no longer need the slave, comment here and then un-assign yourself from the bug. buildduty will take care of it from there.

Comment 34

[Chris Cooper [:coop] (he/him)]

:sfink - this machine is going away as a result of bug 1106922. Do you still need a linux build slave? I can get you one setup in AWS if so?

Comment 35

[Steve Fink [:sfink] [:s:]]

(In reply to Chris Cooper [:coop] from comment #34)
> :sfink - this machine is going away as a result of bug 1106922. Do you still
> need a linux build slave? I can get you one setup in AWS if so?

Ok. I will need one before too long, but I can file a new request at that time (and start squatting again).