I believe there is some issue with MSSQL1.
Monitoring software is showing unbelievable bad performance regarding stability.
Seems to be mostly online, but constantly going offline for 2 or 3 minutes at a time.
Do we have anything serious going on? Are you on the case?
What can we expect?
Thanks
? I have seen no issues at all even on sites using it heavily.
I just checked monitoring, not any outage since this morning, when there was an outage of about 1 hour due to some very bad circumstances and a series of failures, that have been corrected, and SQl reply times have held steady all day long.
Thanks Stephen,
I posted a log from pingdom on Ticket System ( #LNO-98222-202)
Please, if you could check that. I just think pingdom showing those… and together with today´s early issue… was an “odd” coincidence.
Im not sure what to make of it.
It is certainly not down, I have been on your site using it for the last 10-12 minutes, along with another site using it for much of today, and no errors on either one, if it were down you’d certainly have errors.
Thanks for the reassurence.
Jodo people,
It´s really important to have access to server GRAPHS.
Unfortunately, MSSQL1 is one of the servers where graphs are NOT available, (from what I know).
Please, we really need that quality monitoring. And probably other clients too.
Any chance of getting MSSQL1 in theTraffic Grapher ?
Thank you
That traffic grapher wont help you with knowing is a server is up or down, it is simply cpu usage and memory, and the current server simply cant handle it, it start timing out if we add more and makes many timeouts in the monitoring. We have some monitoring from an internal server that gives us this info as well, but it is not available on a public IP. We are working to get such setup through an agreement with another outside datacenter for a better server and network.
These forums are at softlayer, and they have just had a 1 hour outage again for the 5th time this year, we are therfore not going to put the monitoring for public there. We are still workign on the colo agreement for a provider at colo4dallas to mointor such.
Monitoring CPU and memory shouldn’t put that much stress on a server? 0.o
What kind of software are you using?
Maybe I should install it on my ‘other’ database server 
Having trouble getting it to break a sweat at the moment 
http://home.orange.nl/mveldhui/task_manager.png
Granted, that screenshot was taken in the middle of the night, but still..
Well, it has over 400 monitors polling ever 10 to 30 seconds, and over 400 ipchecks monitoring various services.
MRTG/PRTG can all be quite intensive.
(it is a p4 2.4ghz with 1.5GB RAM, that is part of the problem)
Ah ok, on the machine doing the monitoring.. somehow I read that monitoring made MSSQL1 stress out, which didn’t make much sense
Since mssql1 has been upgraded its running great(minus the power strip going off on that port and management interface being down!)
I certainly wasn’t talking about it slowing down with monitoring
Here is a graph, its not 4 cpu HT or 8 core like you have, but its perfectly fine 20% or less load at peak time on shared. (mssql1 server, not monitor server)
http://errorbob.com/mssql1cpugraph.jpg
BTW looks like you could set your cpu affinity better and optimize for all cores, MS CPU License is per socket
In addition if it is per user CAL you can use all available CPUs.
Hmm, a colleague did the actual installation, but we have Processor and I/O affinity checked for all cores. It’s indeed 2 sockets, dual quad core.
When it has more work to do, it starts using more of the cores, but since I rarely log on during peak hours I’ve rarely actually seen 8 cores at work at the same time.
As for HT, this server’s predecessor was a dual CPU HT machine. It got to the point where it would sometimes be at 100% CPU most of the time during the day. We had a lot of problems with “deadlock in communications buffer” and timeouts on the simplest of queries. Updating a single record would sometimes take 3 minutes…
I have never found a full explanation on that problem, but supposedly HT can be partially responsible for deadlocks in SQL Server and some people recommend disabling HT on SQL Server machines.
Yes I too have seen HT issues on SQL, and on FreeBSD for timing issues. HT is not ideal what I wasn’t sure what you had there, the quad cores are pretty sweet But looking forward to the AMD quads now.
in fact any servers i have with SQL server and HT CPU, do not have it enabled due this.