Page built failed: page.waitForFunction: Timeout 30000ms exceeded.
Xuanwo opened this issue · comments
The page has been built and failed since today.
They failed for the same reason three times, so I believe it's not a random error.
Page loaded!
page.waitForFunction: Timeout 30000ms exceeded.
at loadLocalGraph (file:///home/logseq/publish.mjs:74:14)
at async main (file:///home/logseq/publish.mjs:206:5) {
name: 'TimeoutError'
}
Trace file saved to /home/logseq/graph/www/trace.zip
Export failed
Run actions/upload-artifact@v2
With the provided path, there will be 1 file uploaded
Starting artifact upload
The failed actions: https://github.com/Xuanwo/Xuanwo/runs/6021991755?check_suite_focus=true
For friends jumps into this issue:
Quick workaround: use version: 0.6.5
instead.
Let me take a look. Seems the new publish script is still broken somehow.
I found it hard to debug now:
- We can't specify the version of logseq-publish to use (always use
main
) - Historic logseq-publish images are not tagged, we have to use image hash instead
There are three parts that keep changing...
- build scripts
- build image
- logseq itself
Can you please use Logseq version 0.6.5 for now? This should be mostly stable I suppose.
e.g.,
- name: Logseq Publish
# Using main branch to reproduce https://github.com/pengx17/logseq-publish/issues/8
# Should switch back to stable release after this issue fixed.
uses: pengx17/logseq-publish@main
with:
version: 0.6.5
dest: www
Your points are totally valid. We may need a better strategy to handle different versions across script/image/logseq.
I suppose the publish.mjs script should be kept updated with Logseq since the UI selectors may change. hash tag versions should always be available.
I am not sure what about the Action version itself. It seems to be only related to the action.yml
right now, which is mostly never change these days...
Ooops
Line 35 in 1488709
Makes us can't use hashed images like:
FROM ghcr.io/pengx17/logseq-publish@sha256:e6f98c923385fb8207abb3ce313f63cf3ff9c87d8a626cd8d00c322cb2eafe10
This image is the most used version with 8k downloads.
Version 0.6.5 works as expected: https://github.com/Xuanwo/Xuanwo/runs/6022272514?check_suite_focus=true
I think it is still possible by setting it this way:
uses: pengx17/logseq-publish@main
with:
version: "@sha256:e6f98c923385fb8207abb3ce313f63cf3ff9c87d8a626cd8d00c322cb2eafe10"
🤔
Version 0.6.5 works as expected: Xuanwo/Xuanwo/runs/6022272514?check_suite_focus=true
Glad to see it works with the old/stable version. I will replace the master
tag to 0.6.5
for now ...
Anyway, I will take some time working on this fork first to see how to fix the new publish.mjs script
I think it is still possible by setting it this way:
uses: pengx17/logseq-publish@main with: version: "@sha256:e6f98c923385fb8207abb3ce313f63cf3ff9c87d8a626cd8d00c322cb2eafe10"
thinking
Negative.
:) podman pull ghcr.io/pengx17/logseq-publish:@sha256:e6f98c923385fb8207abb3ce313f63cf3ff9c87d8a626cd8d00c322cb2eafe10
Error: invalid reference format
Glad to see it works with the old/stable version. I will replace the
master
tag with0.6.5
for now ...
Thanks for the quick response! Let's use the stable version of logseq as default.
And I will take it a look too when I'm free.
Ops .. yeah, I forgot there is a ":" in the place. shit.
I revised the publish script and published the logseq-publish
with nightly
tag and tried it in this action
https://github.com/pengx17/Xuanwo/actions/runs/2166778252
It seems the new release should work now.
I will hold to stick master to logseq 0.6.5. Will see if there are any failed actions for some time.
I revised the publish script and published the
logseq-publish
withnightly
tag and tried it in this action https://github.com/pengx17/Xuanwo/actions/runs/2166778252It seems the new release should work now. I will hold to stick master to logseq 0.6.5. Will see if there are any failed actions for some time.
I'll give it a try.
I am pushing a new master tag using the latest script. It shows that it is now pretty stable according to https://github.com/pengx17/Xuanwo/actions that there are no failed runs since I changed the waiting logic.