Upgrades, previous containers are not running and upgrade fails because cannot make db dump
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Won't Fix
|
Critical
|
Fuel Python (Deprecated) |
Bug Description
1. install released 5.0 iso
2. run upgrade to 5.0.1 it succeed
3. run upgrade to 5.1
Actual result:
Upgrade script cannot make database dump, because postgresql container was not running
Error: Cannot start container fuel-core-
The reason is during the docker upgrade we update docker and restart the service, docker stops all of the containers, but doesn't umount volumes.
Expected result:
Upgrade succeed
*5.0.1*
VERSION:
mirantis: "yes"
production: "docker"
release: "5.0.1"
api: "1.0"
build_number: "170"
build_id: "2014-08-
astute_sha: "6db5f5031b74e6
fuellib_sha: "a31dbac8fff9cf
ostf_sha: "09b6bccf7d4767
nailgun_sha: "af3d1922bfc213
fuelmain_sha: "fd58828f404e42
*5.1*
VERSION:
feature_groups:
- mirantis
production: "docker"
release: "5.1"
api: "1.0"
build_number: "465"
build_id: "2014-08-
astute_sha: "efe3cb3668b907
fuellib_sha: "52f3ebfa968f03
ostf_sha: "c6ecd0137b5d7c
nailgun_sha: "25eba6fbb2047f
fuelmain_sha: "25a0c228d99870
There are several ways to fix it
1. patch supervisor configs and use dockerctl to run containers, when I tried it last time, we had timeout issues (need to retest it again)
2. stop supervisor and containers before we run docker upgrade
3. before puppet run do docker upgrade explicitly in upgrade script and handle all possible cases there
Personally I like the third solution.