Archive for July 2004
The Alteon load balancers and the PIX firewalls could not be polled using the SNMP poller. At first we suspected the ACL settings, but it turned out that these were correct. The real reason was that the Alteons and the firewalls do not support SNMP version 2. We just switched to version 2 because combined polls are flaky in version 1 (if one poll fails, all polls fail).
Therefore I added an additional configuration file that maps ip addresses to SNMP version numbers in SnmpVersionMapping.conf. The default version is still 2. If you need another version for a device, add an entry to this file.
this solved the problem for the alteons, but not for the PIX firewalls. They still came up empty. I checked it and found that they don’t support the SegmentsIn and SegmentsOut counters. Since they use SNMP version 1 now, this one failed reading caused all readings to fail. So I removed this from the Measurments.conf file. (Note to Stuart, please remove the segments measurements from the firewalls in the XML).
However, this still not resolved the problem. It turned out that the PIX firewalls cannot handle more than 22 combined measurments (they were handling about 30). More than 22 again causes an error. Therefore, I added an additional setting to the SnmpVersionMapping.conf. You can now also specify the maximum number of combined polls per device. I set it to 15 for the PIX’es. If nothing is specified, the default from SnmpPoller.conf is still user.
Everything seems to work now…
Reracking the HP servers was completed. Everything seems to be up and running again after the servers were booted up. One interesting thing though. The Remote insight agents on the HP’s are dependent on then SNMP service, which was stopped to make place for the SSM…
Also, I saw missed heartbeat events for 75 and 77 only. Perhaps the others had cleared, or perhaps they weren’t created. We should investigate.
The urchin server is now up in the development environment. Provisionally it is to be found at ip.ad.dr.ess:9999. login as admin / .
It measures the 4 customer websites . These are staging sites. The equivalent production sites will be up once C installs apache on the two “production” boxes in dev. The staging server is at ip.ad.dr.ess. put entries in your hosts file to the 4 websites above to see them in your browser.
The (under development) CDS system is at ip.ad.dr.ess/cds/login.aspx.
I had a look at single sign for Urchin on from the portal. It looks like in order to do that we need to proxy from the portal, so that Urchin is never directly accesses. We get this for free from the SSL-VPN, so it may be wise to wait with sso until we have the SSL-VPN available for customers.
I tagged our bug fix release of the Cycle 7 tools in CVS as
This is the release that is deployed on the Production servers that are now being moved to S from O.
I tagged the Cycle 6 release of the tools. The tag name is
This is the version deployed on the production servers as of June 11, 2004.
Congratulations and thanks everybody for your hard work!