mirrors/git-annex
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages 1 Wiki Activity Actions

decided not to change this

Browse source
This commit is contained in:
Joey Hess 2025-05-12 12:13:08 -04:00
parent 97203e525b
commit a9eca1b568
No known key found for this signature in database
GPG key ID: DB12DB0FF05F8F38
2 changed files with 24 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

2
doc/todo/cluster_preferred_content_parallel_evaluation_issue_with_archive_group.mdwn
View file

@ -22,3 +22,5 @@ Still it would be good to support this, especially for when existing
archive repositories get put in a cluster.)
--[[Joey]]
> [[wontfix|done]]

22
doc/todo/cluster_preferred_content_parallel_evaluation_issue_with_archive_group/comment_1_74361be79de885bc35cd800a2aad4be0._comment Normal file
View file

@ -0,0 +1,22 @@
[[!comment format=mdwn
username="joey"
subject="""comment 1"""
date="2025-05-12T16:03:36Z"
content="""
If this were implemented, and one of the nodes was full, and that happened
to be the node whose preferred content expression were evaluated first, it
would try to store there and fail, and the content would not be stored in
any of the other nodes. Which seems potentially worse than the file being
stored on multiple nodes currently.
Of course, this is the problem that `sizebalanced` preferred content
solves. Since sizebalanced is otherwise much like the archive group,
it would make sense to just switch the cluster nodes to use it.
As for other preferred content expressions, unless they use
`copies=groupname:number` or `lackingcopies`, whether one node wants
content won't be influenced by what other nodes have it. So evaluating
in preferred content in parallel is ok for those.
I think I've talked myself out of making a change!
"""]]