[09:12] <Chipaca> good morning all
[09:12] <bthomas> morning
[10:18] <facubatista> ¡Muy buenos días a todos!
[10:19] <bthomas> hola
[10:19] <facubatista> hola bthomas
[10:52] <Chipaca> facubatista: mantantirulirulá
[10:52] <facubatista> hola Chipaca :)
[11:07] <mthaddon> Chipaca, facubatista: not sure if you saw as it was a little buried in https://discourse.charmhub.io/t/telegraf-charm-for-k8s/3791 but we have an MP adding functional tests to charm-k8s-telegraf - https://code.launchpad.net/~sajoupa/charm-k8s-telegraf/+git/charm-k8s-telegraf/+merge/393340 - would be interested in your thoughts on this
[11:08] <Chipaca> mthaddon: ack. i'll try to get to it today.
[11:08] <mthaddon> fantastic, thanks
[11:22] <facubatista> mthaddon, thanks for reminding it to us
[11:22] <mthaddon> np
[12:06] <JoseMasson> Good morning!
[12:09] <facubatista> hola JoseMasson
[12:12] <bthomas> Morning JoseMasson : I added replica_set_name to mongodb peer relation data and also ensured that the replicated variable was a string. I pushed my changes to my development branch. Please delete you local copy of development branch before pulling because I did some rebaseing. I would generally not do this but hey it is a development branch :-).
[12:13] <JoseMasson> Great bthomas I'll see it!
[14:08] <mup> PR operator#444 closed: Improve and extend README.md with operator intro <Created by markshuttle> <Merged by chipaca> <https://github.com/canonical/operator/pull/444>
[16:12]  * bthomas -> craving tea
[17:06] <dstathis> bthomas Are any of the charms you have written using hostnames to talk to each other?
[17:06] <dstathis> I'd like to look at some example code
[17:07]  * facubatista -> bbl
[17:07] <bthomas> dstathis: Elasticsearch charm uses hostnames to setup the elasticsearch seed hosts configuration. I am not sure of an example of using that to talk between charms but do not see why not.
[17:09] <bthomas> https://github.com/canonical/elasticsearch-operator
[17:48] <justinclark> bthomas, do you have a few to jump on the Graylog call?
[17:48] <justinclark> Question about mongodb
[17:50] <bthomas> justinclark: going to standup link now
[17:51] <bthomas> justinclark: am there
[17:51] <justinclark> bthomas: https://meet.google.com/jac-ghof-sdn
[18:02]  * Chipaca EODs
[18:02] <Chipaca> ttfn!
[19:23] <bthomas> justinclark: any luck ?
[19:23] <justinclark> bthomas, taking a quick lunch break - I'll let you know in a bit
[19:26]  * facubatista is back
[20:25] <justinclark> bthomas, no luck using the master branch, but it looks like the issue may have something to do with the mongodb defaults. In particular, one log line sticks out to me (trying with standalone mode)
[20:25] <justinclark> 2020-11-10 20:21:21,472 INFO : org.mongodb.driver.cluster - Cluster created with settings {hosts=[mongo:27017], mode=SINGLE, requiredClusterType=UNKNOWN, serverSelectionTimeout='30000 ms', maxWaitQueueSize=5000}
[20:26] <justinclark> The default "hosts=[mongo:27017]" might be the issue because that's what cannot be connected to by Graylog
[20:26] <justinclark> 2020-11-10 20:21:21,605 INFO : org.mongodb.driver.cluster - Exception in monitor thread while connecting to server mongo:27017
[20:26] <justinclark> com.mongodb.MongoSocketException: mongo: Name or service not known
[20:27] <bthomas> There may be multiple problems here. 1) Bind address 2) Mongodb configuration not support slave reads 3) possibly others.
[20:28] <bthomas> If you like I can get on a call with you and JoseMasson now . I will ofcourse look into this first thing tomorrow now that I have a better understanding.
[20:29] <bthomas> justinclark: One thing will help here. Do you know why Graylog is assuming that the name of the mongodb host is "mongo" ?
[20:30] <justinclark> I'm not sure, because we're explicitly setting the standalone/replicaset URI in server.conf
[20:31] <bthomas> justinclark: I am not on my work PC. May I request you to post that replica set URI here.
[20:32] <justinclark> bthomas: directly from the pod's server.conf:
[20:32] <justinclark> mongodb_uri = mongodb://mongodb-0.mongodb-endpoints:27017,mongodb-1.mongodb-endpoints:27017,mongodb-2.mongodb-endpoints:27017/graylog?replicaSet=rs0
[20:33] <bthomas> Hmm... I am surprised . I do not see how Graylog can be picking up the host name of "mongo". Event if it did "mongodb-0" or "mongodb-1" ... I could understand.
[20:34] <justinclark> I'm in the same boat. I'll send you the logs directly.
[20:36] <bthomas> Thanks justinclark : as mentioned above the one thing I worry about is the mongodb bind address. I will dig further tomorrow, and let you have all the fun till then :-)
[20:37] <bthomas> I am online for an hour or two in case I can be of any service but will be winding down, perhaps go watch spitting images https://www.youtube.com/channel/UCCov3Uxwgh6Fq_2F7oW9-hA
[20:47] <bthomas> justinclark: Just a hunch. That if you compare hostnames such as "mongodb-0.mongodb-endpoints" given in the replica set URI with the kind of host names we use in elasticsearch, you may note it is missing a suffix, someting like ".svc. ..." . I wonder if the Graylog is not able to resolve the replica set URI's and falling back to the default "mongo". I guess on can confirm if this is the case by using a shell in the Graylog cont
[20:47] <bthomas> checking if the the host "mongodb-0.mongodb-endpoints" is indeed accessible or not.
[21:25]  * bthomas off for today
[21:32]  * facubatista eods
[21:58] <mup> Issue operator#445 opened: Testing Harness to provide a way to remove relation <Created by ed1000> <https://github.com/canonical/operator/issues/445>