Adar Dembo has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/8352 )

Change subject: tool: new actions for adding and removing data directories
......................................................................


Patch Set 2:

(3 comments)

http://gerrit.cloudera.org:8080/#/c/8352/2/src/kudu/tools/kudu-tool-test.cc
File src/kudu/tools/kudu-tool-test.cc:

http://gerrit.cloudera.org:8080/#/c/8352/2/src/kudu/tools/kudu-tool-test.cc@2330
PS2, Line 2330:   ASSERT_STR_CONTAINS(s.ToString(), "Can't find block");
> shouldn't the tablet fail because it sees that one of its disks is now miss
We do data dir group resolution while loading the tablet's superblock, and hard 
failing there means the tserver will fail to start. I tried this at first (by 
modifying DataDirGroup::FromPB to return a bad status when we can't find a data 
dir) and that's what I got.

Separately, remind me what's the problem with block ID reuse? I know we have a 
problem in tests where we lack the ability to clear the block cache, but what's 
wrong with reuse in production, across distinct tserver process instances?


http://gerrit.cloudera.org:8080/#/c/8352/2/src/kudu/tools/tool_action_fs.cc
File src/kudu/tools/tool_action_fs.cc:

http://gerrit.cloudera.org:8080/#/c/8352/2/src/kudu/tools/tool_action_fs.cc@348
PS2, Line 348: Status AddDataDir(const RunnerContext& context) {
> when adding a data dir, should we have some option to add it to existing ta
That's a good idea, but it'd mean rewriting all the superblocks, so it's not 
trivial to do when you factor in the cleanup work in the event of an error.

Can I punt on it for now?


http://gerrit.cloudera.org:8080/#/c/8352/2/src/kudu/tools/tool_action_fs.cc@554
PS2, Line 554: Any tablets with data on the removed data directory "
             :                         "will fail the next time the server is 
started
> Sort of had a similar thought: perhaps a mandatory unsafe "force" flag if t
I'd rather not introduce interactivity to this particular action, but I'm OK 
with Andrew's idea to require users to pass an "optional" --force parameter. I 
can check for the presence of containers as an approximation for "is there live 
data?", and require users to use --force if there is.



--
To view, visit http://gerrit.cloudera.org:8080/8352
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I6ddbdd6cc6231996e7802a622a8b4691527a0643
Gerrit-Change-Number: 8352
Gerrit-PatchSet: 2
Gerrit-Owner: Adar Dembo <a...@cloudera.com>
Gerrit-Reviewer: Adar Dembo <a...@cloudera.com>
Gerrit-Reviewer: Andrew Wong <aw...@cloudera.com>
Gerrit-Reviewer: Todd Lipcon <t...@apache.org>
Gerrit-Comment-Date: Mon, 23 Oct 2017 23:46:08 +0000
Gerrit-HasComments: Yes

Reply via email to