swap subdirectories disappear with error

[Teyla]

Hello,
Sometimes, it's totally random over time, our webftp upload is stuck at 2Mb with the error in stderr.log :

2021-11-15 02:06:55.444 [STDERR] Failed to create swap file with mkstemp( /tmp/lshttpd/swap/s5/s-G6SlPL ), please check 'Swap Directory' and permission: No such file or directory
2021-11-15 02:06:55.444 [STDERR] Failed to increase temp file size with ftrancate(): Bad file descriptor

and in error.log

2021-11-15 02:06:55.444277 [ERROR] [*****:37766:HTTP2-23] Failed to obtain or reinitialize VMemBuf.
2021-11-15 02:06:55.444376 [ERROR] [****:37766:HTTP2-23#Default] Ran out of swapping space while reading request body!

So, we checked the /tmp/lshttpd/swap, and we figured out that when the problem occurs, we are missing some directories (here by exemple 's5').
If we restart lsws, then we get back the full directory list as :

/tmp/lshttpd/swap# ls
0 1 3 4 5 6 7 8 9 a b c d e s2 s3 s4 s5 s6 s7 s8

The permissions are correct, if i recreate manually the directory, it's working back.

Anyone know why the [s1-s8] directories get randomly deleted ? And how to fix this issue ? We have this issue on different lsws versions, running under debian 9.

Regards

 

[Germont]

In Centos df -h displays a list of partitions and each one used space. Maybe need to increase size of tmp partition.
I don't know the specific command for Debian.