[tahoe-lafs-trac-stream] [Tahoe-LAFS] #475: CPU-watcher munin graph got stuck

[Tahoe-LAFS]

#475: CPU-watcher munin graph got stuck
--------------------------------+------------------------------
     Reporter:  warner          |      Owner:
         Type:  defect          |     Status:  closed
     Priority:  minor           |  Milestone:  undecided
    Component:  code-nodeadmin  |    Version:  1.1.0
   Resolution:  wontfix         |   Keywords:  munin statistics
Launchpad Bug:                  |
--------------------------------+------------------------------
Changes (by exarkun):

 * status:  new => closed
 * resolution:   => wontfix


Old description:

> We had a problem in one of our webapi nodes which caused it to lock up
> (it used a lot of memory, and twistd got an error and tried to kill
> itself, and failed). The node was using 100% CPU for a few minutes.
>
> The problem was that the CPU-watcher kept reporting that 100% CPU to
> munin for the next day and a half (and the cpu percentanges reported for
> the other nodes under its supervision were stuck at their previous values
> too). If the CPU watcher is writing to a file, then we need to change the
> munin plugin to ignore files that are more than 10 minutes old or
> something similar.

New description:

 We had a problem in one of our webapi nodes which caused it to lock up (it
 used a lot of memory, and twistd got an error and tried to kill itself,
 and failed). The node was using 100% CPU for a few minutes.

 The problem was that the CPU-watcher kept reporting that 100% CPU to munin
 for the next day and a half (and the cpu percentanges reported for the
 other nodes under its supervision were stuck at their previous values
 too). If the CPU watcher is writing to a file, then we need to change the
 munin plugin to ignore files that are more than 10 minutes old or
 something similar.

--

Comment:

 Going to delete the current stats code: ticket:3549.

--
Ticket URL: <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/475#comment:2>
Tahoe-LAFS <https://Tahoe-LAFS.org>
secure decentralized storage