5/7/03 : Router, NT5, NT18 and NT19 update.

[admin]

Router should be fixed now. The issue was being caused by the NT5 issue.

NT5 had some kind of IIS worm that would start 10PM and go till 11PM then again at 12am till 4am. This worm should now be removed and should not take down the router or any other part of the network now.


NT18 and NT19 are having ASP issues. There are ALOT of users running bad code on these systems (more then 5 on each). We have kicked more then 20 users off of these 2 systems and we just keep finding more and more problems with users running some kind of bad coding.

We are working to fix these 2 systems but this type of issue we must catch as its happening to fix it.

[kitzco]

The Nt19 is down again You said something about members using bad ASP code. What is bad ASP code and how does this effect the server. Nt19 being down so much is hurting me. I am losing members over the past few days. Why if there is so much trouble with this server is on new signups it still default to this server.

When you say bad code, do you mean people running around doing infinite loops or just small simple mistakes?

[thecomputerpro]

Just as I suspected! I am on NT19 and am noticing ASP go down as well. But as the "Just as I suspected" statement intails, I figured it was someone running bad code and / or abusive code on the system!. I run many Win2000 servers in a corporate environment and use Matrix for my own things. I do not have many if any issues at all on the corporate systems because only I an the web designers have access. Problems like this usually come from the people on the system running something that overloads the server or crashes the environment.

I sure wish those users would test on non-production servers so others like me can keep our production environments running!

I have other customers on other server and they see no issues so I can not imaging that this has anything to do with Matrix and more with the people on those particular machines!

Thanks Admin for the update! Keep finding the abusers and let them know this is not OK!......

Keep up the good work!

[tkraffty]

Just curious - In a case like this, where some user's site is causing problems, do you contact the reseller who owns the user, or the user who owns the site? (or both)

We only contact the reseller. It is up to them to contact their clients. We do shut the site down till they fix the code, then we allow them to turn it on and monitor it from there to make sure it is corrected.

[bootNumlock]

i have myself and two other coders in my little world and we think (for the most part) we write and run some fairly considerate and smart code-- what are some examples of code that is killing the system???

[Brangwyn]

Not destroyting ASP Objects is generally one of the major contributing factors to high resource usage on Windows Servers

eg

Conn.Close()
Set Conn = nothing

I very much doubt Vortech actually look at peoples code before deeming it "bad code" that would take forever to do and would be a waste of their time.

I am actually still slightly concerned about the process used to determine whats "bad code" and what isn't though and would appreciate someone documenting the steps you take during the analysis.

[admin]

Brangwyn, as you said it would take us a very very long time if we went over every ones code. We can watch whats going on with the server and whos code is running for long times with out closing. We take note of who is doing it each time after we see that domain 5 times in the list of running way to long say over 1,000,000 ms something is wrong..

At that point we stop the site in IIS let the reseller know so they can let there customer know something must be wrong with there code or something in there site. NT19 and NT18 seems a little better now but still want it to be more stable so still trying to find the rest of the problems..

NT20 is about done now, we had some hardware issue and had to send them back for RMA, but also have NT21 and NT22 about done just waiting on some of there parts..

I hope to get all 3 online by monday at least NT20 by the end of this week if not sooner..

[Brangwyn]

That sounds pretty fair Brad, I was a little worried that using iistracer you'd just see a site hung and mark it down as being bad, giving it 5 shots is pretty fair it may occasionally give you a red herring if a site is a very high volume one simply becuase its most likely to have requests outstanding or underway that could hang if something stops working on the server.

One thing you might want to throw into the mix is looking at what state the hung process is in, if its in the same process each time theres an extremely likely chance it is indeed crap code, if its different each time then I'd be a little more cautious personally before calling it bad.

I guess if all else fails though shutting a site off and seeing if the server stops missbehaving is a pretty foolproof method too

Great news about the additional servers, as always great service