Project

General

Profile

Fail cases recovery

Julia Zashchitina
Added almost 4 years ago

Hello. We had a quick database disconnect and after that MUC component doesn't work properly:

2015-04-23 13:39:57.593 [in_0-message-router] MessageRouter.processPacket() FINEST: Processing packet: from=sess-man@ip-172-31-12-104.us-west-2.compute.internal, to=516969a746bc4d0686ae0fe9af7ac6da@muc.domain.com, DATA=Service not available., SIZE=391, XMLNS=jabber:client, PRIORITY=NORMAL, PERMISSION=NONE, TYPE=error

Could you, please, suggest:

  1. What is a standard recovery way for situations like this?

  2. Can we automatically restart Tigase server in general fail cases?

  3. Is it possible to reload separate components (MUC for example) without restarting Tigase server?


Replies (5)

Avatar?id=6023&size=32x32

Added by Artur Hefczyc TigaseTeam almost 4 years ago

Tigase has a logic for automated reconnecting to the DB if it discovers that a connection was broken. We tested it many times and it seems to be working OK. There is no need for any manual actions or restarting either the whole Tigase server or a component.

However, it appears that it did not work in your case for some reason. We would need to run an investigation on the problem to find out why Tigase recovery mechanisms did not work. If you have Tigase logs from that time, maybe there is some information which would allow us to pinpoint why it did not recover from DB connection failure automatically. If you have some details on the case, please submit a bug report with all information that you have:

  1. Tigase server version

  2. Database you use

  3. Database driver that you use

  4. OS

  5. Logs,

  6. Tigase configuration

Any additional information you may have.

Added by Julia Zashchitina almost 4 years ago

Thanks for your reply. Unfortunately, we do not have Tigase logs for that period any more. As for another information that you requested:

1) Tigase server version: 7.0.0

2) Database: Amazon RDS with MySQL 5.6

3) Database driver: MySQL Connector Java version 5.1.35

4) OS: Ubuntu 12.04 LTS

5) Tigase configuration:

--comp-class-1 = tigase.muc.MUCComponent
--virt-hosts = domain.com,ip-xxx-xx-xx-xxx.us-west-2.compute.internal
--user-db-uri = jdbc:mysql://host/tigasedb?user=user&password=password
--user-db = mysql
--admins = admin@domain.com
--comp-name-4 = message-archive
--comp-name-3 = proxy
config-type = --gen-config-all
--comp-name-2 = pubsub
--comp-name-1 = muc
--cluster-mode = true
--sm-plugins = +message,-amp,+msgofflinepush,-msgoffline
--debug = server
--comp-class-4 = tigase.archive.MessageArchiveComponent
--comp-class-3 = tigase.socks5.Socks5ProxyComponent
--comp-class-2 = tigase.pubsub.PubSubComponent

proxy/component-id=proxy@domain.com
muc/message-filter-enabled[B]=false

--comp-name-5=http
--comp-class-5=tigase.http.HttpMessageReceiver

--vhost-register-enabled=false
--api-keys=open_access

If you have any opened issue for this case, please, post a link.

Avatar?id=6023&size=32x32

Added by Artur Hefczyc TigaseTeam almost 4 years ago

I have created a bug report for this issue: #3041. The bug report is in the MUC component as it is most likely specific to this code.

Added by Julia Zashchitina almost 4 years ago

We have faced the issue again. Unfortunately, we only have a console log (please, see in attachment).

Added by Julia Zashchitina over 3 years ago

MUC-fail issue has repeated again after database disconnect.

    (1-5/5)