I have an EC2 instance (Ubuntu 20) I took management over. It has been fine for the last year or so. There is an S3 bucket mounted in fstab like so :
s3fs#bucket-name /path/to/mount fuse _netdev,url=https://s3-ca-central-1.amazonaws.com,endpoint=ca-central-1,allow_other 0 0
The instance had this mounted and was working fine. Yesterday, the instance stooped responding, I had to restart the instance using the EC2 console. After that, the bucket (in fact both of them) don't mount. When I issue a command like df or mount, I get :
df: /path/to/mount: Transport endpoint is not connected.
I tried mounting the bucket manually with s3fs in debug :
s3fs -d -d -f -o url="https://s3-ca-central-1.amazonaws.com" -o endpoint=ca-central-1 -o allow_other bucket-name /path/to/mount
I get :
[CRT] s3fs.cpp:set_s3fs_log_level(297): change debug level from [CRT] to [INF]
[INF] s3fs.cpp:set_mountpoint_attribute(4400): PROC(uid=0, gid=0) - MountPoint(uid=0, gid=0, mode=40775)
And then it sits there indefinitively with no output.
The instance has also an IAM role that grants permission to that bucket. Doing "aws s3" commands on it works fine.
/etc/passwd-s3fs has the correct credentials and permissions.
Launching a new instance with ubuntu 20, installing s3fs and creating /etc/passwd-s3fs file with the same content works just fine.
Any hint on how to debug this ?
Thanks for that. AFAIK, the s3fs cache option isn't there. So from what I understand from the man page, cache should not be used. Is there another cache that could be involved ?