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

Linux executable binary files read corruption #357

Closed
bhzhu203 opened this issue Feb 21, 2022 · 9 comments
Closed

Linux executable binary files read corruption #357

bhzhu203 opened this issue Feb 21, 2022 · 9 comments

Comments

@bhzhu203
Copy link

bhzhu203 commented Feb 21, 2022

many execute binary files runing failed in the bcachefs disk in version 7d09250
The cinnamon desktop Enviroment startup error and some settings seems to be error

'''
./firefox
bash: ./firefox: cannot execute binary file: Exec format error
'''

mounting the filesystem
'''
[ 4291.839048] bcachefs (00000000c0634e7a): recovering from clean shutdown, journal seq 2255293
[ 4296.217905] bcachefs (00000000c0634e7a): going read-write
[ 4296.219710] bcachefs (00000000c0634e7a): error creating gc thread: -4
[ 4296.219712] bcachefs (00000000c0634e7a): error starting gc thread
[ 4322.681470] bcachefs (000000004b1ee5ab): recovering from unclean shutdown
[ 4468.111361] bcachefs (000000004b1ee5ab): journal read done, 0 keys in 1 entries, seq 2255295
[ 4472.524304] bcachefs (000000004b1ee5ab): journal replay done
[ 4532.672503] bcachefs (000000004b1ee5ab): going read-write
[ 4532.681538] bcachefs (000000004b1ee5ab): mounted with opts: compression=lz4,foreground_target=cache,background_target=back,promote_target=prom,noinodes_use_key_cache,nojournal_transaction_names
[ 4532.681936] mount (10380) used greatest stack depth: 10168 bytes left
'''

when umount the filesystem:
bcachefs (000000004b1ee5ab): error 1 from bch2_move_data() in copygc

@bhzhu203
Copy link
Author

bhzhu203 commented Feb 21, 2022

updated to 9d554fa still the same .
many files read corruption

‘’‘
bhzhu@bhzhuos /m/b/s/b/bcachefs> git status
error: bad signature 0x00000000
fatal: index file corrupt
bash-4.3$ /mnt/bhzhu/gocryptfs
bash: /mnt/bhzhu/gocryptfs: cannot execute binary file: Exec format error
bash-4.3$ /mnt/bhzhu/firefox/firefox
bash: /mnt/bhzhu/firefox/firefox: cannot execute binary file: Exec format error
'''

'''
[13270.312400] bcachefs (000000001ded2859): error 1 from bch2_move_data() in copygc
[13326.364768] bcachefs (000000007504deec): recovering from clean shutdown, journal seq 2258168
[13330.768120] bcachefs (000000007504deec): going read-write
[13330.770051] bcachefs (000000007504deec): mounted with opts: compression=lz4,foreground_target=cache,background_target=back,promote_target=prom,noinodes_use_key_cache,nojournal_transaction_names
'''

version f76e97e can read files Correctly ,but the panic issue exists #352

@bhzhu203 bhzhu203 changed the title Warning! Linux execute binary files read corruption in 7d0925084b6927ad8c631bde92fb1c41cc6270c2 Linux executable binary files read corruption in 7d0925084b6927ad8c631bde92fb1c41cc6270c2 Feb 22, 2022
@bhzhu203 bhzhu203 changed the title Linux executable binary files read corruption in 7d0925084b6927ad8c631bde92fb1c41cc6270c2 Linux executable binary files read corruption Feb 22, 2022
@bhzhu203
Copy link
Author

"Linux executable binary files read corruption" issue solved in 79b1c2e

I start format a brand new bcachefs , a single device , open option compress:lz4 ; backgrand_compress:lz4

using rsync copy my btrfs home directory data (3000000+ files , 176G ) to bcachefs directory , when reaching 80G bcachefs , the whole system panic . Everytime I want to continue rsync , the whole system panic in 5mins again and again

@koverstreet
Copy link
Owner

Did you get the log?

@bhzhu203
Copy link
Author

bhzhu203 commented Mar 9, 2022

Good news : I have update to version e318fab , then I can do the disk rsync copy completely. And I have using bcachefs for several days , there is no panic issue .

More stable now. But the "bch-rebalance/s" process often runing in the backgroud ,sometimes laging. The good thing is the "bch-rebalance/s" not using full of the disk read bandwidth

@bhzhu203
Copy link
Author

bhzhu203 commented Mar 9, 2022

Unfortunately,after 2 hours I was writing the feedback , The system panic.

I had deleted tons of small files :

rm -rvf ~/.cache/fontconfig/*

deleting faild ,out of the list.So I using the rsync to delete

rsync -avP --delete ~/null/ ~/.cache/fontconfig/

At this time, the system was not painc yet.

Finally I search the bookmark in the firefox,trying poping out a new window,the whole system panic.

Maybe firefox stores many small files in the disk. Deleting and creating often.

Then I reboot the system ,mounting the disk ,when entering the Cinannom desktop , the system panic immediately.
I have to mount the disk in other system , let the bcachefs "bch-rebalance/s" runing completed (iotop -o showing no active process).

The FS becomes normal now.

@koverstreet
Copy link
Owner

Are you using snapshots? I just fixed several transaction path overflows when using snapshots

@bhzhu203
Copy link
Author

Are you using snapshots? I just fixed several transaction path overflows when using snapshots

Yes I am using snapshots. @koverstreet

This version is 62cdb94 now. Quite stable when doing rsync and using Firefox browser . I have keept/test this version nearly for a week , not experienced any panic issue this version yet.

But I find some performance issues:

  1. When the disk mounted , "bch-rebalance/s" process will run for about 5mins. At this time ,the desktop is laging. The good thing is that "bch-rebalance/s" seems be limited at disk bandwidth , not runing out of the disk read bandwidth after hours using.

2.Bcachefs performance drops than before. Virtualbox Windows10 vm on Bcachefs:

tested at 2021.11.15
as-ssd-bench VBOX HARDDISK 2021 11 15 9-44-49

tested at 2022.3.3
as-ssd-bench VBOX HARDDISK 2022 3 3 17-48-41

I will continue test more snapshots on the latest version .Thanks

@koverstreet
Copy link
Owner

Was this actually fixed?

@bhzhu203
Copy link
Author

bhzhu203 commented May 16, 2022

Was this actually fixed?

Yes, It is more stable now. using the version=new_data_types . But there is a issue that slow/low HDD disk random read IOPS with many snapshots

#387
@koverstreet

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

No branches or pull requests

2 participants