Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
Kristen
Test
22859 Posts |
 Posted - 2007-02-08 : 07:38:17
|
| We have two boxes - IIS and SQL.We are getting periodic very slow service from IIS.I can get a static HTML url no problem, but an ASP url, which goes to the database, runs slow.It runs fine for 10 minutes or so, then slows down again.There is no significant CPU, Memory or Disk activity on either box (as per PerfMon).There are 1,300 TCP connections open on IIS box (as reported by NETSTATS), 900 of these are to the SQL box - mostly saying "TIME_WAIT"Seems a lot!! but I don't have a baseline for this particular statistics.Any ideas?Thanks, Kristen |
|
|
eyechart
Master Smack Fu Yak Hacker
3575 Posts |
Posted - 2007-02-08 : 11:53:20
|
| The first thing I would look at are you ethernet adapters and network switches. Make sure that they are set to 100/FULL and not auto negotiate (if your LAN supports this). If they are gigabit adapters they should be set to auto negotiate. ALso, if you have gig adapters in your servers, but your network infrastructure is not gigabit, then you may have to do some tweaking of the switch settings. The auto negotiate capabilities in 100mbit stuff are crap and they were implemented slightly differently by each vendor. In gigabit, the auto negotiate capability is a part of the spec so each vendor interoperates properly.This is usually called link flapping, and if this problem is occuring you should be able to see evidence of this in the logs on your switches and routers.-ec |
 |
|
|
Kristen
Test
22859 Posts |
Posted - 2007-02-08 : 12:05:21
|
| Thanks for that ec.We have our own switch at the ISP - so its only got a couple of boxes to worry about. They are all Gigabit, and have been running fine for ages - so whilst I doubt the NIC settings could have been changed I'll double check that someone didn't do an accidental-click while using TS. I'll have a word with the hardware bods to check, or swap, the cards etc.I hadn't though about logs in the switches, I'll look into that.We've rebooted the IIS box (but it was previously rebooted yesterday afternoon - a bit of a short timeframe to get big memory leaks etc.) and throttled the site that had the most activity and that seems to have calmed things down, but its all a bit weird given that nothing appeared to be maxing out.Kristen |
|
|
|
eyechart
Master Smack Fu Yak Hacker
3575 Posts |
Posted - 2007-02-08 : 13:25:55
|
| NICs can go bad, switch modules go bad, switches actually go bad and cables can get damaged. All of those can cause this kind of thing.I would just thoroughly check the network to make sure that it is all working as it should.-ec |
|
|
|
Kristen
Test
22859 Posts |
Posted - 2007-02-08 : 14:14:53
|
Excellent! Its a 4 hours drive (that's a LONG WAY for us Brits!) and we have about 0.000001" of snow (that means that Brit Land has ground to a halt) ..... now then, how shall I send?  |
|
|
|
eyechart
Master Smack Fu Yak Hacker
3575 Posts |
Posted - 2007-02-08 : 14:48:26
|
| You can do much of the checking remotely. Try pinging hosts using large packets and see if you get any drops. Look through logs on servers and switches/routers. Use tools like iPerf to check throughput http://dast.nlanr.net/Projects/Iperf/-ec |
|
|
|
|
|
|
|
|
Server connection slow