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

Daily nightwatch transfer #21

Merged
merged 5 commits into from Dec 4, 2020
Merged

Daily nightwatch transfer #21

merged 5 commits into from Dec 4, 2020

Conversation

weaverba137
Copy link
Member

This PR fixes #20.

In addition, adds an ensure_scratch() function so that if ${CSCRATCH} is unavailable, an alternate temporary directory can be chosen.

I'm still testing the new rsync commands and exclude files, but it looks good so far.

@weaverba137 weaverba137 self-assigned this Nov 20, 2020
Copy link
Contributor

@sbailey sbailey left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good. I especially like the automated alternative if $CSCRATCH isn't available. Confirming: does any big data flow through that, or only logs and/or small temporary files? i.e. do we need to worry about the small $HOME quota? If anything big (like actual data files) flows through that, consider using a CFS dir as the scratch alternative instead.

@weaverba137
Copy link
Member Author

Yes, the scratch directory is only used for setting up temporary files used by the morning "catch-up" transfers and by the HPSS backups. No actual data lands there.

@weaverba137
Copy link
Member Author

@sbailey, while trying to do a full re-sync of spectro/nightwatch/kpno, I noticed a lot of files are now missing from KPNO. Is that because at some point you went from individual spectrograph files, to files containing data from all spectrographs? Or did entire classes of files become obsolete? For example, 20200308/00054444/qfiberflat-b0-00054444.fits is gone.

@sbailey
Copy link
Contributor

sbailey commented Nov 20, 2020

Ah, yes, I removed additional semi-optional files from KPNO to clean up disk space for new data. This will be an ongoing "feature" of nightwatch: occasionally removing a subset of files while leaving the most important ones. Eventually Klaus is going to do that to the raw data too, removing directories that previously were there.

Could you configure the NERSC rsync to retain whatever files it had previously successfully synced, without generating N>>1 false alerts about files that disappear on the KPNO side?

If you let me know when you do the daily sync, I can coordinate that any cleanup won't be happening at that time so that we don't remove files out from under you during a sync.

@weaverba137
Copy link
Member Author

The current default for the daily transfer is to not delete anything on the NERSC side (i.e. rsync is run without the --delete option). So that's not a problem.

The daily transfer is initiated at 12:00 MST.

@weaverba137
Copy link
Member Author

Daily nightwatch transfers are working fine, & I want to get some other minor changes into the tag, so I'll merge this as soon as tests pass.

@weaverba137 weaverba137 merged commit 030dfdc into master Dec 4, 2020
@weaverba137 weaverba137 deleted the nightwatch-engineering branch December 4, 2020 00:36
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.

Shift nightwatch transfer to daily engineering data transfer
2 participants