Hello,
We’re getting a problem on one of our web
servers.
Here’s the
set-up:
Web
Server:
Dell Power Edge
2450
Win2k
SP2
IIS
5
Norton Antivirus – set-up to avoid the
scanning global.asa, (as opening the global.asa can causing site to stop
responding)
Database
Server
Dell Power Edge
2450
Win2k SP1 (I’m applying SP2
shortly)
SQL
2000
The web server has a direct connection to
the web with no firewall. The database server connects to the web server via a
secondary network card in the web server and on a separate IP network and
hence hiding the DB server from the outside
world.
The web server has 3 sites on it all
running different versions of the same site with 3 different versions of our
custom built ActiveX component. The 3 sites include - one Live, one customer
trial site and one in-house demo/dev site. The customer is testing new
developments on the customer trial site, and their customers are using the
live site.
The sites work fine and then after a while
they will slow and eventually stop responding. The server requires a full restart to
get it working again, Inetinfo sits near the bottom of the process list not
really doing anything and not using a great deal of memory. I can’t see anything in the process
list that is battering the servers’
resources.
We have regressed any new developments we
put into the trial site before these problems started to happen, in case it
was any strange code making the server
crash.
The last weird thing that happened before
the sites started playing up, was the web server reported an out of virtual
memory error. The websites
stopped working, the server then reallocated memory then came back to life,
allowing users to access the sites again. One thing we noticed after this is
that all dates changed to American format rather than United Kingdom format –
so we changed these back and restarted the server for good
measure.
Looking through the Event Viewer there is
nothing too unusual in there apart from an error saying the SSL is screwed, it
really says: “A fatal error occurred while creating an SSL server credential”,
although the client has a certificate protected site it is not used. Another error we seem to be getting is
script timeouts, but these don’t correlate with server
crashes.
We do get lots of CodeRed hack attempts but
I patched the server when the hotfix first came out from
MS.
Does anybody have any ideas? As I am running out of
them.
Thanks
Paul
p.s. if the client had a bad connection to
the server that dropped regularly, (due to their overloaded network) could
that cause a server not to clear sessions and “get confused” and eventually
fall over – I’m clutching at straws
now.
Pps.
A virtual luxury apartment anywhere in the world that you want with a
virtual keg of beer to anyone that
helps.
Paul Broomfield, Network Administrator and Database
Technician
Tell
Communications
Tamar Science
Park
Derriford
Plymouth
Devon
PL6
8BX
Tel: +44 (0)
1752 764242
Fax: +44 (0)
1752 764243
Visit us on
the web at: http://www.tell-com.com/
http://www.sunbelt-software.com/ntsysadmin_list_charter.htm