Description
Increasingly we are seeing both technical and administrative issues with needing to run cbcollect_info independently on each node and then manually uploading it to our servers.
The problems range from:
-Running the command on more than one node causes problems
-Inconsistent naming and uploading problems causes increased back-and-forth between customer and support
-Increased administrative overhead on the customer for larger and/or multiple clusters
I'd like to get discussion going on a proposal to make this better for our customers and support:
The end goal would be to have a UI link and a CLI command that can be used to automatically generate a cbcollect_info from all (or just one) node and then have those/that file(s) automatically uploaded to a default (or specified) location.
Requirements would look like:
1) An internal process to generate a cbcolect_info on one or more nodes. If more nodes, the process should prevent the tool from running on more than one node at a time. It should take as input a user supplied cluster name, the hostname(s) of the nodes it is running on and a location on disk (default to /tmp, allow for overriding). It should return status of completion on each node, and a location on disk where the file has been created.
2) Either a separate process or part of the above one should also allow for uploading the file to a specified location. The default would be our S3 bucket, but also allow for an FTP or HTTP location to be supplied if required (due to changing locations and individual environment restrictions such as firewalls). It should return status of completion for each node, and the location where the data was uploaded to.
3) Either tool should allow for no more than 3 cbcollect_info's to be stored on disk at any time and automatically clean them up to preserve disk space.
4) Tooling:
a) REST API
b) CLI
c) UI: Propose replacing the "generate diagnostic report" link in the logs page of the UI with a pop-up that asks the customer to:
-choose a specific node or all nodes
-override the default location for upload (S3) to specify an FTP or HTTP location
-override the default location on disk (/tmp) to specify an envrionment specific path
-Pressing "ok" would trigger the internal processes above
Intermediate steps could include just 1 and 4b above, with 2, 3 and 4a/4c to come later.
Attachments
Issue Links
- depends on
-
MB-10086 Cluster-wide diagnostics gathering - collect_info from UI across cluster
- Closed