Closed jvshahid closed 6 years ago
We have created an issue in Pivotal Tracker to manage this:
https://www.pivotaltracker.com/story/show/153368219
The labels on this github issue will be updated when the story is started.
/cc @sunjayBhatia @aminjam
https://github.com/cloudfoundry-incubator/consul-release/releases/tag/v191 contains a property for this, I could consider that it closes your issue, but I also want your feedback on what you find to be a sane default value for that property (i.e. what makes WATS pass reliably) so I will leave open for now.
Made a PR for the windows-cell ops-file ^ and the value can be changed based on WATS needs.
Context
Historically we have disabled the windows Dnscache service because it was caching negative results (i.e. failure to resolve a name) which caused long outages if windows tried to query consul agent while it was down (e.g. while consul agent is restarting). We later realized that the consul agent isn't very stable and would sometimes fail to resolve names (which is something the Diego team is experiencing). @mhoran mentioned that the current solution is to enable the Dnscache service. Set
MaxNegativeCacheTtl
to0
in order to prevent caching of failures and rely on Dnscache service to cache entries so that we don't have to rely on the consul agent being very reliableI am not sure if the caching of
.service.cf.internal
entries has been verified or if that's a recent regression. I am inclined to say it is the former since consul docs indicates that the default TTL is set to 0 to prevent caching. You can also check the ttl of the record returned from consul by runningExpected bevavior
when i run
I expect the output of
ipconfig /displaydns
to include a record forbbs.service.cf.internal
to indicate that the entry is cachedActual behavior
I do not see any
.cf.internal
entries in the dns cache