Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[b4.4] Fix "fill" handling in static decomposition #1500

Merged
merged 4 commits into from
Nov 12, 2024

Conversation

narategithub
Copy link
Collaborator

This PR addressed #1495 . This PR (with #1498 and #1499) has been tested with these test cases in ldms-test repo (branch b4.4):

Narate Taerat added 2 commits November 5, 2024 09:24
If "fill" was correctly specified, `cfg_col->fill` will have a value.
Otherwise, `cfg-col->fill` will be `NULL`. We should use `cfg_col->fill`
to determine if "fill" was present.
The "fill" string value was handled incorrectly. This patch fixes the
bug.
@narategithub narategithub changed the title Fix "fill" handling in static decomposition [b4.4] Fix "fill" handling in static decomposition Nov 5, 2024
@morrone
Copy link
Collaborator

morrone commented Nov 7, 2024

Still broken, but in a different way:

ldmsd: decomp_static.c:1014: resolve_col: Assertion `mlen' failed.

Error message isn't terribly helpful. I am not sure what it isn't liking.

@narategithub
Copy link
Collaborator Author

@morrone can I see your config?

@narategithub
Copy link
Collaborator Author

@morrone never mind me. It was an incorrect assertion that mlen should not be 0 for all types. The length should be ignored for non-array types. The fix is on the way ...

@narategithub narategithub marked this pull request as draft November 7, 2024 14:47
@narategithub narategithub added this to the v4.4.5 milestone Nov 8, 2024
Narate Taerat added 2 commits November 7, 2024 21:04
- Adress the issue that the static decomposition incorrectly set
  `LDMS_V_LIST` as type of the column when "type" was not specified in
  the config.

- In the case of multiple lists with different length in a set, the
  number of produced rows equals to the number of the longest list. When
  the shorter lists are exhausted, the "fill" value of the respective
  column is used. However, if "fill" was not specified, `ldmsd` crashed
  due to `NULL` fill mval. This patch uses a `zfill` (zero value) as a
  fall back if "fill" was not specified.
@narategithub narategithub marked this pull request as ready for review November 8, 2024 03:26
@narategithub
Copy link
Collaborator Author

@morrone Could you please give this another whirl?

@morrone
Copy link
Collaborator

morrone commented Nov 9, 2024

@narategithub It looks like there are no longer any errors thrown with my config and your latest fix. My testbed is slightly broken (networking issues) at the moment, so I can't quite check the full decomposition output for correctness. But I can see that my "cluster" default fill is at least working, so that is a good sign.

@narategithub
Copy link
Collaborator Author

Thank you, @morrone !

@tom95858 tom95858 merged commit b33f9d0 into ovis-hpc:b4.4 Nov 12, 2024
14 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants