Thursday, June 02, 2005

If you're really bored... You can follow my luck with the MS newsgroups...

I've got a few background issues at the moment:

  • One is Performance Monitor on my Win2k3 Dev box.
    It's gone nutz and lost all the performance objects (CPU, Index Server, Disk, etc, etc) and only displays numbers...
    Someone has previously reported this, but got no response and their post has mysteriously disappeared from the MS Web Newsgroup but thankfully still appears in Google. [Actually, and I thought I had, just found that I posted to the groups a while back on this and still no response, as an MSDN Universal subscriber, I'm supposed to get a response within 48 hours from MS or an MVP...]
  • Another is with the ISA2004 install procedure for setting up SQL as the logging back-end. [MS, Google]
    Looks like there's a mistake in the docs...
  • An older one is based around an ISAPI filter for IIS, to do user impersonation based on client IP address, but I'm still waiting on the MVP to write up his demo code.
Another issue I have, which I haven't posted yet, since I'm still trying to narrow it down, is with Index Server on Win2k3.
Just a couple of weeks ago, it went ballistic and decided that it needed to index millions of documents in directories that only had a few thousand docs.
I think I've narrowed it down to a performance setting. If you set Indexing to "instant" and there are more documents than it can scan in a given interval (haven't found where that interval is documented yet), then they just start piling up... And it gets worse if you have multiple catalos... I have three. One (System) has around 200,000 docs to worry about, another (custom for an app) has about 100,000 and the third is the standard Web one, with only about 1000 docs.
Why this happened all of a sudden I don't know and I'm sure I didn't have the Indexing set to "instant" when the problem started. I changed it to that to try and get through the indexing quicker! This seems to be an old problem [link above is for 2002], so I would have thought people would have hit it more often and MS would have fixed it so that the same document doesn't appear in the "Docs to Index" queue more than once, or for re-indexing not to get queued until the last one completes.


If I could get Performance Monitor working, I'd check out the disk activity, since CPU/RAM usage is low (on this Dual XEON 2.4GHz HyperThreaded 2Gig RAM, 2*36Gig SCSI [RAID 1], 3*72Gig [RAID 5]) it must be disk holding it back, but it didn't use to be a problem, all three catalos have been around for months and the size hasn't changed much. Oh well, I'm pausing all the catalos and then letting them index one at a time, when that's done, I'll change the performance setting (back to lazy) and hopefully they'll be happy... If not [yells]MICROSOFT![/yells]


Oh and just remembered anotherone that's in the background.
We run SharePoint Team Services v1.0 (aka 2001 I think) and it's got a quirk (pre OfficeXP sp3) of not being able to delete users once the user table get's fairly large, due to too many foreign keys. (infact SQL goes nuts and lets everything fall on the floor...)
I've followed the proceedure on the support page, twice now and it still does it...

Right now I've got it going off doing a "Server Health Check", which entails matching up all the files and DB records, making sure that everything makes sense... Which promptly got it sending out Change Notification emails [on my behalf] for things that haven't changed... I guess they were changes that were queued up in the system somewhere... [sigh] scared the heck out of some of our users when they thought I was messing with documents [I work for a lawfirm remember...]
Note to self: don't let MS products do "auto repair" functions, just get them to detect and report them. Then I can pick up the pieces without scaring the users...

No comments: