I recently resolved a performance problem by stopping and restarting the http cluster resource on our E2K cluster. I've been trying to figure out ever since why http would affect the server's performance.
Around 9:15a users started complaining that e-mail "seemed slow." I confirmed that messages were sitting in the Outbox for up to four minutes, and that other tasks like opening large folders and deleting messages were taking much longer than expected. After checking the usual suspects (nothing of note in the event log; cpu utilization was around 20%; comm between the cluster nodes was fine), I went into ESM and saw that the SMTP local delivery queue was holding between 20 and 40 messages. I traced a test message to myself and saw that it took five minutes to travel through the queue. It's rare for our queue to exceed 1, so that confirmed to me that something was wrong. After about an hour of fruitless snooping around, I sent a firmwide e-mail and tried stopping and restarting the SMTP cluster resource. That took forever (well, about 10 min) but when it was done the queue remained high. Next I tried the MTA cluster resource, but again, no luck. Then I restarted the HTTP cluster resource and the queue emptied out almost instantly. Users also immediately reported better performance. I know E2K works closely with IIS (ExIPC, e.g.) but IIS and HTTP are obviously not the identical, so I'm unsure why restarting HTTP would have such a dramatic effect. I've checked my E2K books, as well as technet and winnetmag.com, but http exchange 2000 is not exactly a narrow query. Any thoughts? E2K SP3 on W2K SP2. _________________________________________________________________ List posting FAQ: http://www.swinc.com/resource/exch_faq.htm Archives: http://www.swynk.com/sitesearch/search.asp To unsubscribe: mailto:[EMAIL PROTECTED] Exchange List admin: [EMAIL PROTECTED]
