Forums

Home / Forums

You need to log in to create posts and topics. Login · Register

docker-tar images


minimos
6 Posts
May 7, 2025, 7:33 am

We have a PetaSAN 3.1 system (which will be upgrade soon, but that's another story), with 3 nodes that offers NFS and iSCSI services.
Perhaps due to a recent pg inconsistency, while logging on the nodes to issue some ceph repairs commands I noticed that in /var/tmp there was an accumulation of files named docker-tar<numericstring> that was almost filling the filesystem.
Apparently they were caused by some podman tasks related to petasan-nfs-ganesha ?

My question is why they were accumulated and not wiped away after podman ran its course?
I found a few traces of that in syslog, but not enough to understand what happened, e.g. if the container aborted or failed.

#1

minimos
6 Posts
May 7, 2025, 11:53 am

Update: from /opt/petasan/log/PetaSAN.log there are some sections like:

07/05/2025 14:40:31 INFO Starting NFS Exports Service
07/05/2025 14:40:31 INFO Container Manager : creating NFS-10-1-3-2 container
07/05/2025 14:40:32 ERROR Error running cmd : podman run --name NFS-10-1-3-2 --hostname NFS-10-1-3-2 -p 10.1.3.2:2049:2049 -d -
t localhost/petasan-nfs-ganesha:3.2.0
07/05/2025 14:40:32 ERROR Error: cannot listen on the TCP port: listen tcp4 10.1.3.2:2049: bind: cannot assign requested address

07/05/2025 14:40:32 ERROR NFSExportsServer : init_action() : Cannot create and run new container "NFS-10-1-3-2"
07/05/2025 14:40:32 INFO NFSServer : clean local resource : NFS-10-1-3-2
07/05/2025 14:40:32 INFO NFSServer : clean local resource : 1) Stopping NFS Exports service of resource : NFS-10-1-3-2
07/05/2025 14:40:32 INFO NFSServer : clean local resource : 2) Delete ip address of resource : NFS-10-1-3-2
07/05/2025 14:40:32 ERROR LockBase : start() --> Error during process.
07/05/2025 14:40:32 ERROR expected str, bytes or os.PathLike object, not NoneType
Traceback (most recent call last):
File "/usr/lib/python3/dist-packages/PetaSAN/core/consul/lock_base.py", line 264, in start
self.process()
File "/usr/lib/python3/dist-packages/PetaSAN/core/consul/lock_base.py", line 321, in process
while not self.do_process(): # if it returns False , try again
File "/usr/lib/python3/dist-packages/PetaSAN/core/consul/lock_base.py", line 348, in do_process
self.clean_mismatched_local_resources()
File "/usr/lib/python3/dist-packages/PetaSAN/core/consul/lock_base.py", line 549, in clean_mismatched_local_resources
self.clean_local_resource(local_resource)
File "/usr/lib/python3/dist-packages/PetaSAN/core/nfs/nfs_server.py", line 904, in clean_local_resource
ret = NetworkAPI().delete_ip(ip=str(ip), eth=self.eth_name, subnet=str(subnet))
File "/usr/lib/python3/dist-packages/PetaSAN/core/network/network.py", line 70, in delete_ip
p = subprocess.Popen(["ip", "address", "del", "/".join([ip, subnet]), "dev", eth], stdout=subprocess.PIPE,
File "/usr/lib/python3.8/subprocess.py", line 858, in __init__
self._execute_child(args, executable, preexec_fn, close_fds,
File "/usr/lib/python3.8/subprocess.py", line 1639, in _execute_child
self.pid = _posixsubprocess.fork_exec(
TypeError: expected str, bytes or os.PathLike object, not NoneType
07/05/2025 14:40:32 INFO Clean all local resources.

We use the range 10.1.3.1-10.1.3.3 for the NFS, and all the 3 IPs are assigned to one of the 3 nodes (not sure how often they move from one node to another)

#2