Clustering Tools (NETVIRT-1002)

[NETVIRT-1003] Tool to identify leader and follower nodes Created: 17/Nov/17  Updated: 25/Nov/19

Status: Open
Project: netvirt
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: Sub-task Priority: Medium
Reporter: Sam Hague Assignee: Tim Rozet
Resolution: Unresolved Votes: 0
Labels: csit:3node
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Blocks
is blocked by GENIUS-138 Improve Datastore Cluster diagstatus ... Resolved

 Description   

The CSIT tests need a method to capture the current follower and leader nodes. The information should be dumped to the karaf log and/or a file so that it is easy to reference the state along with the accompanying test. The tool would be called to collect the status at that instant of time and then logged. The tests would call the tool periodically to collect the status.

The monitor tool currently displays the leader and follower nodes, but it runs continuously and is not applicable to a running csit.

 Comments   
Comment by Michael Vorburger [ 25/Jun/18 ]

I'm not crystal clear about what exactly the requirement is here (a "tool" as in? a REST URL? CLI command? with what input and output, specifically?) but it seems to me that GENIUS-138 may either already answer this, or could perhaps be a suitable basis for building upon.

Comment by Sam Hague [ 25/Jun/18 ]

The requirement doesn't care how the information is retrieved - we just needed a tool that could grab the data and then we could dump it. The cluster monitor tool that is python-based and does rest calls to odl to request the data. That tool is a cli tool that runs continuously and updates status and does not capture a single status.

We would still need a tool to retrieve the diagstatus in GNEIUS-138. Other question on GENIUS-138, is will it show leader and follower status at all times for all nodes? There is a comment that says error status is reported if not all status are right. The tool we need will be used in CSIT and live debugging and needs to show the leader or follower status for all nodes in the cluster.

Comment by Michael Vorburger [ 26/Jun/18 ]

> The cluster monitor tool that is python-based and does rest calls to odl to request the data.
> That tool is a cli tool that runs continuously and updates status and does not capture a single status.

OK, so what is wrong with that tool - is it not sufficient? Or is this JIRA just about using that tool in CSIT? Or are we looking for something similar to that tool but built into ODL as a REST URL or CLI command instead of an external python-based one?

>  question on GENIUS-138, is will it show leader and follower status at all times for all nodes? There is a comment that says error status is reported if not all status are right. The tool we need will be used in CSIT and live debugging and needs to show the leader or follower status for all nodes in the cluster.

OK; you're right - GENIUS-138 is a tool who's primary goal is more to return a boolean true/false "all good or not" feedback - not provide details about leader and follower status (although it does something like that in the diagstatus description in case of an error). But it internally does have that information, so if you would like to have something built which prints that e.g. from a new CLI command (like... "mdsal-cluster:status" kind of thing in controller, perhaps?), then that should be possible.

Comment by Tim Rozet [ 26/Jun/18 ]

vorburger the tool currently prints to a gui terminal. I'll add an option to log changes only to a file. That's all we need, no need to expose new cli in ODL imo

Comment by Abhinav Gupta [ 25/Nov/19 ]

any update here?

Generated at Wed Feb 07 20:22:59 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.