[9fans] silent fossil crashes

[9fans] silent fossil crashes

[mirtchov]

it happened again -- fossil just stopped working without any sign of
error. no sysfatal, no crash, nothing.  just stopped responding.

just when it died I had run a script that pulls from sources. the script does:

	echo 'srv -AWP somethingsomething' >> /srv/fscons

at which point fossil decided to take a break.

i have a saved a serial console log of attempting to run ls, showing
that the machine isn't responding and then doing ^T^Tp and ^T^Tq
before the reboot.  if anyone is interested.

^T^Tq didn't show any processes waiting in the queue, whish wasn't the
case with previous (nrdy>=1) crashes that were fixed a few weeks ago.

^T^Tk killed one of the fossil processes promptly.

hope that helps: andrey

Re: [9fans] silent fossil crashes

[David Presotto]

[-- Attachment #1: Type: text/plain, Size: 58 bytes --]

How many tcp connections are open to that fossil (if any)?

[-- Attachment #2: Type: message/rfc822, Size: 2362 bytes --]

From: mirtchov@cpsc.ucalgary.ca
To: 9fans@cse.psu.edu
Subject: [9fans] silent fossil crashes
Date: Tue, 2 Dec 2003 09:51:37 -0700
Message-ID: <c7e1e0281acf33d0d51f8cb7df42b5c1@plan9.ucalgary.ca>

it happened again -- fossil just stopped working without any sign of
error. no sysfatal, no crash, nothing.  just stopped responding.

just when it died I had run a script that pulls from sources. the script does:

	echo 'srv -AWP somethingsomething' >> /srv/fscons

at which point fossil decided to take a break.

i have a saved a serial console log of attempting to run ls, showing
that the machine isn't responding and then doing ^T^Tp and ^T^Tq
before the reboot.  if anyone is interested.

^T^Tq didn't show any processes waiting in the queue, whish wasn't the
case with previous (nrdy>=1) crashes that were fixed a few weeks ago.

^T^Tk killed one of the fossil processes promptly.

hope that helps: andrey

Re: [9fans] silent fossil crashes

[David Presotto]

[-- Attachment #1: Type: text/plain, Size: 83 bytes --]

Also (if you can) keep a 'con /srv/fscons' open somewhere
to see error messages...

[-- Attachment #2: Type: message/rfc822, Size: 2362 bytes --]

From: mirtchov@cpsc.ucalgary.ca
To: 9fans@cse.psu.edu
Subject: [9fans] silent fossil crashes
Date: Tue, 2 Dec 2003 09:51:37 -0700
Message-ID: <c7e1e0281acf33d0d51f8cb7df42b5c1@plan9.ucalgary.ca>

it happened again -- fossil just stopped working without any sign of
error. no sysfatal, no crash, nothing.  just stopped responding.

just when it died I had run a script that pulls from sources. the script does:

	echo 'srv -AWP somethingsomething' >> /srv/fscons

at which point fossil decided to take a break.

i have a saved a serial console log of attempting to run ls, showing
that the machine isn't responding and then doing ^T^Tp and ^T^Tq
before the reboot.  if anyone is interested.

^T^Tq didn't show any processes waiting in the queue, whish wasn't the
case with previous (nrdy>=1) crashes that were fixed a few weeks ago.

^T^Tk killed one of the fossil processes promptly.

hope that helps: andrey

Re: [9fans] silent fossil crashes

[mirtchov]

> Also (if you can) keep a 'con /srv/fscons' open somewhere
> to see error messages...

will do.  i use cons to log one of the consoles, but that's only as
good as looking the last command that crashed:

	Tue Dec  2 04:03:28: main: archive vac:xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
	Tue Dec  2 09:10:27: srv -AWP somethingsomething
	Tue Dec  2 09:28:00: prompt:

you can see the machine being rebooted after srv -AWP...

andrey

Re: [9fans] silent fossil crashes

[mirtchov]

> How many tcp connections are open to that fossil (if any)?

not too many..  i had two clients connected to it.  the total number
of fossil processes at the time of crash were 23, so about 15-20 tcp
connections.

Re: [9fans] silent fossil crashes

[Russ Cox]

> > How many tcp connections are open to that fossil (if any)?
>
> not too many..  i had two clients connected to it.  the total number
> of fossil processes at the time of crash were 23, so about 15-20 tcp
> connections.

i'd be interested to see the ^t^tp list.
also the output of "msg" on your current console.

Re: [9fans] silent fossil crashes

[jmk]

On Tue Dec  2 13:50:34 EST 2003, rsc@swtch.com wrote:
> > > How many tcp connections are open to that fossil (if any)?
> >
> > not too many..  i had two clients connected to it.  the total number
> > of fossil processes at the time of crash were 23, so about 15-20 tcp
> > connections.
>
> i'd be interested to see the ^t^tp list.
> also the output of "msg" on your current console.

and the output of "srv" on the fossil console.