action #159888
closedOpen pull requests on Backlog Limits Checker use up too much space
0%
Description
Observation¶
From #158236#note-36:
Open pull requests seem to include files from other open pull requests, eventually causing errors suggesting there is no space file:
error: unable to write file pr-preview/pr-47/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-45/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-43/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-41/pr-preview/pr-41/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-41/index.html
fatal: cannot create directory at 'pr-preview/pr-47/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-45/pr-preview/pr-46/pr-preview/pr-46/pr-preview/pr-43/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-41/pr-preview/pr-41/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-42/pr-preview/pr-41/pr-preview': No space left on device
Depending on what's open you can see it on the github-pages branch.
Suggestions¶
- Consider more recent changes related to checking out branches, e.g. https://github.com/openSUSE/backlogger/pull/29
Files
Updated by livdywan 8 months ago
- Copied from action #158236: Backlog Limits Checker github workflow fails on pull requests from forks size:S added
Updated by livdywan 8 months ago
I have a draft: https://github.com/openSUSE/backlogger/pull/50
One idea I have is that checking out a second copy for state.json ends up being pulled into the update later. So that could be remedied via https://github.com/actions/checkout#fetch-only-a-single-file using sparse-checkout: state.json; sparse-checkout-cone-mode: false or by using a different folder name that isn't re-used.
Unfortunately it seems my changes are never executed. Neither are they on my own fork https://github.com/kalikiana/backlogger/pull/3 where there's also something else that breaks... so maybe the fix must be merged untested?
Updated by livdywan 8 months ago
- Status changed from In Progress to Resolved
Unfortunately it seems my changes are never executed. Neither are they on my own fork https://github.com/kalikiana/backlogger/pull/3 where there's also something else that breaks... so maybe the fix must be merged untested?
After getting the change reviewed, I went ahead and merged it and successfully tested it with a throwaway branch. No more unwanted files!
Updated by tinita 8 months ago
- Status changed from Resolved to Workable
We still get recursive directories:
...
* f874035 2024-05-02 Remove preview for PR 49 🛬 [kalikiana]|
| pr-preview/pr-49/index.html | 86 ----------------------------------------------
| pr-preview/pr-49/pr-preview/pr-11/index.html | 71 --------------------------------------
| pr-preview/pr-49/pr-preview/pr-19/index.html | 77 -----------------------------------------
| pr-preview/pr-49/pr-preview/pr-20/index.html | 71 --------------------------------------
| pr-preview/pr-49/pr-preview/pr-49/index.html | 86 ----------------------------------------------
| pr-preview/pr-49/pr-preview/pr-49/pr-preview/pr-11/index.html | 71 --------------------------------------
| pr-preview/pr-49/pr-preview/pr-49/pr-preview/pr-19/index.html | 77 -----------------------------------------
| pr-preview/pr-49/pr-preview/pr-49/pr-preview/pr-20/index.html | 71 --------------------------------------
| pr-preview/pr-49/pr-preview/pr-49/preview.png | Bin 10858 -> 0 bytes
| pr-preview/pr-49/pr-preview/pr-49/state.json | 1 -
| pr-preview/pr-49/preview.png | Bin 10880 -> 0 bytes
| pr-preview/pr-49/state.json | 1 -
| 12 files changed, 612 deletions(-)
* b49bd7e 2024-05-02 Deploy preview for PR 49 🛫 [kalikiana]|
| pr-preview/pr-49/index.html | 4 +--
| pr-preview/pr-49/pr-preview/pr-49/index.html | 86 ++++++++++++++++++++++++++++++++++++++++++++++
| pr-preview/pr-49/pr-preview/pr-49/pr-preview/pr-11/index.html | 71 ++++++++++++++++++++++++++++++++++++++
| pr-preview/pr-49/pr-preview/pr-49/pr-preview/pr-19/index.html | 77 +++++++++++++++++++++++++++++++++++++++++
| pr-preview/pr-49/pr-preview/pr-49/pr-preview/pr-20/index.html | 71 ++++++++++++++++++++++++++++++++++++++
| pr-preview/pr-49/pr-preview/pr-49/preview.png | Bin 0 -> 10858 bytes
| pr-preview/pr-49/pr-preview/pr-49/state.json | 1 +
| pr-preview/pr-49/preview.png | Bin 10858 -> 10880 bytes
| pr-preview/pr-49/state.json | 2 +-
| 9 files changed, 309 insertions(+), 3 deletions(-)
* 4a0634b 2024-05-02 Deploy preview for PR 49 🛫 [kalikiana]|
| pr-preview/pr-49/index.html | 86 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
| pr-preview/pr-49/pr-preview/pr-11/index.html | 71 ++++++++++++++++++++++++++++++++++++++++++++++++++++
| pr-preview/pr-49/pr-preview/pr-19/index.html | 77 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
| pr-preview/pr-49/pr-preview/pr-20/index.html | 71 ++++++++++++++++++++++++++++++++++++++++++++++++++++
| pr-preview/pr-49/preview.png | Bin 0 -> 10858 bytes
| pr-preview/pr-49/state.json | 1 +
| 6 files changed, 306 insertions(+)
Updated by tinita 8 months ago
I did clean up the gh-pages branch now, though.
There were still leftover files for older pull requests:
[gh-pages e3a70aa] Remove older PR previews
3 files changed, 219 deletions(-)
delete mode 100644 pr-preview/pr-11/index.html
delete mode 100644 pr-preview/pr-19/index.html
delete mode 100644 pr-preview/pr-20/index.html
Last week I only reset the branch to commit ca0d3eb like I mentioned in slack, I did not do a cleanup.
It was just meant as a first aid to be able to actually check out the branch without getting a full disk.
Updated by livdywan 8 months ago
tinita wrote in #note-6:
Last week I only reset the branch to commit ca0d3eb like I mentioned in slack, I did not do a cleanup.
It was just meant as a first aid to be able to actually check out the branch without getting a full disk.
Right. I couldn't see it in the history. As far as I could confirm all branches created recently were cleaned up correctly by the workflow, so this must have been left-overs from another issue.
Also, thanks!