lessfs crashes with "Database write failed : Cannot allocate memory"

[Maciej Piechotka]

Feb 06 16:55:40 mpiechotka-dt-l lib_bdb.c[4356]: Exit function : bdb_lock : 1486428940:750194
Feb 06 16:55:40 mpiechotka-dt-l lib_bdb.c[4356]: Entering function : release_bdb_lock : 1486428940:750469
Feb 06 16:55:40 mpiechotka-dt-l lib_bdb.c[4356]: Exit function : release_bdb_lock : 1486428940:750483
Feb 06 16:55:40 mpiechotka-dt-l lib_bdb.c[4356]: Exit function : bin_write_dbdata : 1486428940:750628
Feb 06 16:55:40 mpiechotka-dt-l file_io.c[4356]: Exit function : file_update_inuse : 1486428940:750725
Feb 06 16:55:40 mpiechotka-dt-l lib_common.c[4356]: Entering function : get_hash_lock : 1486428940:750834
Feb 06 16:55:40 mpiechotka-dt-l lib_common.c[4356]: Entering function : get_hash_lock : 1486428940:750938
Feb 06 16:55:40 mpiechotka-dt-l lib_common.c[4356]: Exit function : get_hash_lock : 1486428940:751040
Feb 06 16:55:40 mpiechotka-dt-l lib_common.c[4356]: Entering function : release_hash_lock : 1486428940:751164
Feb 06 16:55:40 mpiechotka-dt-l lib_common.c[4356]: Exit function : release_hash_lock : 1486428940:751360
Feb 06 16:55:40 mpiechotka-dt-l lib_bdb.c[4356]: Entering function : bin_write_dbdata : 1486428940:751477
Feb 06 16:55:40 mpiechotka-dt-l lib_bdb.c[4356]: Entering function : bdb_lock : 1486428940:751581
Feb 06 16:55:40 mpiechotka-dt-l lib_bdb.c[4356]: bdb_lock : bin_write_dbdata
Feb 06 16:55:40 mpiechotka-dt-l lib_bdb.c[4356]: bdb_lockedby : bin_write_dbdata
Feb 06 16:55:40 mpiechotka-dt-l lib_bdb.c[4356]: Exit function : bdb_lock : 1486428940:751894
Feb 06 16:55:54 mpiechotka-dt-l lib_bdb.c[4356]: bin_write_dbdata : database 2 keylen 16 datalen 24
Feb 06 16:55:54 mpiechotka-dt-l lib_bdb.c[4356]: Database write failed : Cannot allocate memory

I tried to reduce max threads to 1 but it didn't helped. I'm using lessfs from Gentoo portage.

Maciej Piechotka

[Maciej Piechotka]

Found that I should've probably set TUNEFORSIZE (though currently the actual data is <1TB). Note that with chunks_io it resulted in corruption as data file does not exists.

[Maciej Piechotka]

Unfortunatly for medium it is still reproducable and for large I got a out of memory error.

[Emmanuel Florac]

How much memory do you have? If you are low on memory adding swap can help.

--
You received this message because you are subscribed to the Google Groups "lessfs " group.
To unsubscribe from this group and stop receiving emails from it, send an email to lessfs+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Maciej Piechotka]

16 GiB of main memory, 4 GiB of zram, 16 GiB of swap. 4 GiB is consumed by non-locked KVM though. In total I have 8 GiB of available memory (after lessfs crash). I wouldn't think about it as particularly memory constraint system.

Data is backed by btrfs/lvm/(bcache + RAID 10) so I don't think throughput should be slow.

To unsubscribe from this group and stop receiving emails from it, send an email to lessfs+un...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to a topic in the Google Groups "lessfs " group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/lessfs/XXGwzjhPR-A/unsubscribe.
To unsubscribe from this group and all its topics, send an email to lessfs+un...@googlegroups.com.

[Maciej Piechotka]

I forgot to add - swap is mostly empty so in total lessfs should have about 24 GiB of memory to commit.