-
Notifications
You must be signed in to change notification settings - Fork 851
[BUG] another coredump for version 3.4.6-1 #9302
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
Comments
Here's yet another coredump:
|
And another:
|
|
Could you plz analyze and fix these crashes ? |
Hi @daoxian, I've had a look at the last backtrace (#9302 (comment)), and it looks broken. Would it be possible for you to provide us with the coredump? Best, Tobias |
Thank you for your quick response. Do you mean the coredump files? |
Yes, exactly. |
OK, wait for hours please, for the files are very large. |
@goedderz I've uploaded relating coredump files: Thanks! |
Hi @daoxian, thanks a lot for sharing. I am currently downloading the files. We will have a look and then come back to you. |
Hi @daoxian,
I'm also calculating SHA256 sums, but this is going to take a few hours. |
Yes they're the same size as those on my local machine. |
@goedderz Everyday I can see new coredump files. Two of them listed as follows:
If you need coredump files, I can also supply. |
Thanks a lot. For now we're fine and probably have enough information to fix it, otherwise we will ask you for more. Best, Tobias |
Thanks, and please let me know when this issue gets fixed |
now I am curious did you enable the collection cache ? |
No, the startup cmd is: |
Btw, there're over 10,000 tables in one database. And all the tables are created dynamically. |
Hi @graetzer, I think I've found the coredump reason: After bulk import, I called collection.unload() method, while pregel on this collection is running. This is related to the coredump info "getCollectionIdLocal". |
Hi @graetzer , I think you've fixed that, which version or git branch should I use to verify this fix ? |
Thanks for your patch. But I've found new coredumps after disable unload. Maybe my previous guess was wrong.
|
is there a chance you are using empty edge collection names ? |
Do you mean an edge collection that doen't exists or an empty edge collection containing no edges? An empty edge collection is very likely to be used in my codes, this is a common case. And when an exception occurs while creating a collection, there IS a chance that the subsequent codes using a graph whose edge collection doesn't exist. |
Btw, can vertex or edge collection be modified (adding or removing vertex/edge) when pregel is running on that graph? |
This is yet another occurrence of the same issue as in #9302 (comment) and #9301 (comment) As already mentioned, this looks very much like a race on |
@dothebart Is there any update on the core dump mentioned in - #9302 (comment) I am facing a similar core in the prod -
|
3.4.6-1 , CentOS-7.4, Single mode on metal machine.
The text was updated successfully, but these errors were encountered: