Hi,
On 2014-09-23 07:45, Nathan Scott wrote:
>
>> The only real concern I have at the moment is the need for the
>> query_interval/timestamp stuff which is pretty ugly but when doing e.g.
>> "pminfo -dfmtT ds389.userroot" two successive queries are done for some
>> reason, I think we should avoid such needless LDAP queries.
>
> You may find this is as a result of the query "batchsize" concept within
> pminfo - see the -b,--batch option. If this is indeed the cause, you'll
> find the fetch routines are passed sets of metrics, & the PMDA can respond
> to just those requested (as this one does, to an extent, with use of the
> PMID cluster field). Specific sub-trees can be queried via ds389.userroot
> or ds389.cn and you should see only those LDAP queries fire off.
thanks, the batchsize explains why it happens only with ds389.userroot
not with ds389.dn (the former has > 20 metrics, the latter < 20). With
-b 100 only one query. So I think the query interval logic still has its
place.
> Yes, we'll definitely want to be doing that :) & in an automated fashion.
> A test just like the memcached test qa/835 will be needed - any interest in
> tackling that to finish this PMDA off?
I'll try to take a look at it in the coming days but please feel finish
it off before that :)
> I found the perl-LDAP package is also needed - the Install script could be
> tweaked to check for this. We'll also need to stitch this into the build
> via a new makefile, and assign a domain number. I'll take on these last
> few, but would love some testing help if you can. If not, I'll come back
> to that as soon as I can too. For now, I've merged in your initial patch
> so that we have a common starting point (not built/installed yet though).
Looks good, thanks a lot.
Cheers,
--
Marko Myllynen
|