[check_postgres] RFC: New Actions

[David E. Wheeler]

On Jan 7, 2011, at 6:33 PM, Greg Sabino Mullane wrote:

>> * Idle transaction count: Report error if more 
>> than x idle transactions more than x [time units] old.
> 
> Should be quite do-able, and I think people have even asked 
> about this before (Steven or Robert maybe?). Not sure if 
> this would be shoehorned into the existing txn_idle action 
> with some additional flags or its own action. I'd lean towards 
> the former, especially as I suspect those x/x combos may be 
> handy for other future actions (so keep the arg names genericy 
> please :)

Ah, yeah, good thinking. Since --warning and --critical are times, perhaps add --warning-count and --critical-count for the numbers? So

    check_postgres_txn_idle --warning='15 seconds' --warning-count=10

Would trigger a warning if there were 10 or more transactions that had been idle for more than 15 seconds. Have a look at the attached patch and let me know what you think. A native French speaker will need to review my Google-translate-powered translations, which I'm sure are just awful.

>> * Wal Log Timeout: Report if no new logs on a slave 
>> within the previous x [time units].

We actually realized that this is easily done with the core Nagios file system tools. Duh.

One other thing we wanted to do was to refine the bloat check query. Right now it's an either/or situation, which can result in a lot of false positives. So what we'd like to be able to do is to specify *both* a minimum bloat size *and* a percentage. Thoughts?

Best,

David

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Add-critical-count-and-warning-count-options-for-txn.patch
Type: application/octet-stream
Size: 6601 bytes
Desc: not available
Url : https://mail.endcrypt.com/pipermail/check_postgres/attachments/20110114/b76c06ca/attachment.obj