Reminder, disk is getting low

Reminder, disk is getting low

[Jason Molenda]

The /sourceware/www disk is 90% full, it's got ~500-600MB of space left.
I know there are plans to migrate to a new box - I hope it happens in the
rather near future or mail archiving will have some new problems. :-)

Red Alert (was: Reminder, disk is getting low)

[Gerald Pfeifer]

On Mon, 15 Jan 2001, Jason Molenda wrote:
> The /sourceware/www disk is 90% full, it's got ~500-600MB of space left.
> I know there are plans to migrate to a new box - I hope it happens in the
> rather near future or mail archiving will have some new problems. :-)

  [gerald@sourceware ~]$ df -k /www
  Filesystem           1k-blocks      Used Available Use% Mounted on
  /dev/sdc3              5654061   5120306    240971  96% /sourceware/www

This is getting incredibly tight and getting urgent.

Can we hook another disk to that box and move that file system over? What
else?

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: Red Alert (was: Reminder, disk is getting low)

[Jason Molenda]

On Fri, Feb 02, 2001 at 12:16:20AM +0100, Gerald Pfeifer wrote:
> On Mon, 15 Jan 2001, Jason Molenda wrote:
> > The /sourceware/www disk is 90% full, it's got ~500-600MB of space left.

>   [gerald@sourceware ~]$ df -k /www
>   Filesystem           1k-blocks      Used Available Use% Mounted on
>   /dev/sdc3              5654061   5120306    240971  96% /sourceware/www


It seems unlikely that 250MB disappeared in two weeks.  I'd bet
there was something other than the normal archiving of mail messages,
web pages, and httpd access logs that ate the space.

J

Re: Red Alert (was: Reminder, disk is getting low)

[Jason Molenda]

On Thu, Feb 01, 2001 at 03:22:48PM -0800, Jason Molenda wrote:

> It seems unlikely that 250MB disappeared in two weeks.  

It may be cyclical.  I see that this week's gcc access_log and
sourceware access_log combined are about 200MB.  They'll be scrolled
back and compressed on Saturday (~36 hours from now).  My note on
the 15th noting 500MB of free space was sent on a Monday, so this
variability may just be the normal ebb and flow of the log files.

In any event, the free space is awfully low - I'd agree with Gerald
that this should be addressed.  It'd be an annoying hassle for the
current sourceware maintainers if this disk filled.  If people are
looking for a short-term fix, move the old, compressed log files
to another filesystem.  (I'd council against deleting the access
logs - you never know when you want to show the traffic levels over
a long time period.  Old error log files are rarely useful; IMHO
they could be removed).

Jason

Re: Red Alert (was: Reminder, disk is getting low)

[Jeffrey A Law]

  In message < 20010201152817.A20922@shell17.ba.best.com >you write:
   > In any event, the free space is awfully low - I'd agree with Gerald
  > that this should be addressed.  
It's on order and will probably be installed in early March on my next
planned trip to the Sunnyvale area.  It's a stopgap until we've got the
new box and move into the co-loc.



jeff

URGENT: disk full (was: Red Alert)

[Gerald Pfeifer]

On Tue, 20 Feb 2001, Jeffrey A Law wrote:
>> In any event, the free space is awfully low - I'd agree with Gerald
>> that this should be addressed.
> It's on order and will probably be installed in early March on my next
> planned trip to the Sunnyvale area.  It's a stopgap until we've got the
> new box and move into the co-loc.

Folks, this is getting *way* too close:

  /dev/sdc3  5654061  5343156  18121 100% /sourceware/www

Especially considering that none has been able to regenerate those web
mail archives that have been broken (for whatever reason), we simply
cannot allow a disk full situation which also breaks that archiving
completely.

If the situation has not improved by tomorrow morning (GMT+0100), I'll
try to look for some emergency fixes, but I'd really prefer someone else
having a look.

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: URGENT: disk full (was: Red Alert)

[Christopher Faylor]

On Fri, Mar 09, 2001 at 07:54:00PM +0100, Gerald Pfeifer wrote:
>On Tue, 20 Feb 2001, Jeffrey A Law wrote:
>>> In any event, the free space is awfully low - I'd agree with Gerald
>>> that this should be addressed.
>> It's on order and will probably be installed in early March on my next
>> planned trip to the Sunnyvale area.  It's a stopgap until we've got the
>> new box and move into the co-loc.
>
>Folks, this is getting *way* too close:
>
>  /dev/sdc3  5654061  5343156  18121 100% /sourceware/www
>
>Especially considering that none has been able to regenerate those web
>mail archives that have been broken (for whatever reason), we simply
>cannot allow a disk full situation which also breaks that archiving
>completely.
>
>If the situation has not improved by tomorrow morning (GMT+0100), I'll
>try to look for some emergency fixes, but I'd really prefer someone else
>having a look.

I am looking at this now.  This is a little perplexing since I just deleted
some relatively big files yesterday.  I confess that I didn't check the
disk usage after I did this, though.

cgf

Re: URGENT: disk full (was: Red Alert)

[Jeffrey A Law]

  In message < Pine.BSF.4.33.0103091949480.24504-100000@deneb.dbai.tuwien.ac.at >
you write:
  > On Tue, 20 Feb 2001, Jeffrey A Law wrote:
  > >> In any event, the free space is awfully low - I'd agree with Gerald
  > >> that this should be addressed.
  > > It's on order and will probably be installed in early March on my next
  > > planned trip to the Sunnyvale area.  It's a stopgap until we've got the
  > > new box and move into the co-loc.
  > 
  > Folks, this is getting *way* too close:
  > 
  >   /dev/sdc3  5654061  5343156  18121 100% /sourceware/www
  > 
  > Especially considering that none has been able to regenerate those web
  > mail archives that have been broken (for whatever reason), we simply
  > cannot allow a disk full situation which also breaks that archiving
  > completely.
  > 
  > If the situation has not improved by tomorrow morning (GMT+0100), I'll
  > try to look for some emergency fixes, but I'd really prefer someone else
  > having a look.
The new disk is scheduled to go in Mon or Tues.
jeff

Re: URGENT: disk full (was: Red Alert)

[Jason Molenda]

On Fri, Mar 09, 2001 at 01:57:23PM -0500, Christopher Faylor wrote:

> I am looking at this now.  This is a little perplexing since I just deleted
> some relatively big files yesterday.  I confess that I didn't check the
> disk usage after I did this, though.
> 

It's Friday, the biggest day of the log files.  It really does make
a difference.  It'll be better tomorrow.  Until Friday next week, anyway. :)

-rw-r--r--   1 webuser  webuser  157246098 Mar  9 11:33 sourceware-access_log
-rw-r--r--   1 webuser  webuser  58872893 Mar  9 11:33 gcc-access_log
-rw-r--r--   1 webuser  webuser   1972244 Mar  9 11:33 gcc-error_log
-rw-r--r--   1 webuser  webuser   5145844 Mar  9 11:32 sourceware-error_log
-rw-r--r--   1 root     webuser  16054795 Mar  9 11:31 ha.redhat.com-access_log
-rw-rw-r--   1 root     webuser    253013 Mar  9 10:51 ha.redhat.com-error_log


That 157MB sourceware access log file will get a 10:1 compression factor
tonight when it's bzipped.  Same for the other logs.


You can copy some of the older log files off to another disk/system.
I'd keep them around until you get more disk space on the box -
it's occasionally useful to examine old log files to look at growth
over time, and it's a shame to lose them after they've been kept
all this time.


(On the other hand, IMHO it's not really useful to keep around the old
error logs from 1998/1999/2000.  I can't remember _ever_ looking at an
old error log.  I don't know if this would get you much space savings,
though)

Jason

Reminder, disk is getting low

[Jason Molenda]

The /sourceware/www disk is 90% full, it's got ~500-600MB of space left.
I know there are plans to migrate to a new box - I hope it happens in the
rather near future or mail archiving will have some new problems. :-)

Re: URGENT: disk full (was: Red Alert)

[Christopher Faylor]

On Fri, Mar 09, 2001 at 07:54:00PM +0100, Gerald Pfeifer wrote:
>On Tue, 20 Feb 2001, Jeffrey A Law wrote:
>>> In any event, the free space is awfully low - I'd agree with Gerald
>>> that this should be addressed.
>> It's on order and will probably be installed in early March on my next
>> planned trip to the Sunnyvale area.  It's a stopgap until we've got the
>> new box and move into the co-loc.
>
>Folks, this is getting *way* too close:
>
>  /dev/sdc3  5654061  5343156  18121 100% /sourceware/www
>
>Especially considering that none has been able to regenerate those web
>mail archives that have been broken (for whatever reason), we simply
>cannot allow a disk full situation which also breaks that archiving
>completely.
>
>If the situation has not improved by tomorrow morning (GMT+0100), I'll
>try to look for some emergency fixes, but I'd really prefer someone else
>having a look.

I am looking at this now.  This is a little perplexing since I just deleted
some relatively big files yesterday.  I confess that I didn't check the
disk usage after I did this, though.

cgf

Re: Red Alert (was: Reminder, disk is getting low)

[Jason Molenda]

On Thu, Feb 01, 2001 at 03:22:48PM -0800, Jason Molenda wrote:

> It seems unlikely that 250MB disappeared in two weeks.  

It may be cyclical.  I see that this week's gcc access_log and
sourceware access_log combined are about 200MB.  They'll be scrolled
back and compressed on Saturday (~36 hours from now).  My note on
the 15th noting 500MB of free space was sent on a Monday, so this
variability may just be the normal ebb and flow of the log files.

In any event, the free space is awfully low - I'd agree with Gerald
that this should be addressed.  It'd be an annoying hassle for the
current sourceware maintainers if this disk filled.  If people are
looking for a short-term fix, move the old, compressed log files
to another filesystem.  (I'd council against deleting the access
logs - you never know when you want to show the traffic levels over
a long time period.  Old error log files are rarely useful; IMHO
they could be removed).

Jason

Re: Red Alert (was: Reminder, disk is getting low)

[Jason Molenda]

On Fri, Feb 02, 2001 at 12:16:20AM +0100, Gerald Pfeifer wrote:
> On Mon, 15 Jan 2001, Jason Molenda wrote:
> > The /sourceware/www disk is 90% full, it's got ~500-600MB of space left.

>   [gerald@sourceware ~]$ df -k /www
>   Filesystem           1k-blocks      Used Available Use% Mounted on
>   /dev/sdc3              5654061   5120306    240971  96% /sourceware/www


It seems unlikely that 250MB disappeared in two weeks.  I'd bet
there was something other than the normal archiving of mail messages,
web pages, and httpd access logs that ate the space.

J

Red Alert (was: Reminder, disk is getting low)

[Gerald Pfeifer]

On Mon, 15 Jan 2001, Jason Molenda wrote:
> The /sourceware/www disk is 90% full, it's got ~500-600MB of space left.
> I know there are plans to migrate to a new box - I hope it happens in the
> rather near future or mail archiving will have some new problems. :-)

  [gerald@sourceware ~]$ df -k /www
  Filesystem           1k-blocks      Used Available Use% Mounted on
  /dev/sdc3              5654061   5120306    240971  96% /sourceware/www

This is getting incredibly tight and getting urgent.

Can we hook another disk to that box and move that file system over? What
else?

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/