You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
mount-s3 bucket --prefix prefix /data/s3/prefix --max-threads=1 --debug --log-directory /var/log/mount-s3
Same behavior observed if executed with the required options only
mount-s3 bucket /data/s3/prefix
What happened?
Separate mount of BTRFS on /data - EBS/instance store, separate for each instance
S3 is mounted on /data/s3/prefix, using a specific prefix on a bucket
Initial S3 mount from user data on instances is successful
For faster restart of our services, we need to format mount on /data device, meaning that we need to unmount /data/s3/prefix, format /data, and then remount /data/s3/prefix back
After remounting the S3 mountpoint, we get a message that it was successful:
2024-01-09T12:09:52.754068Z INFO mount_s3: successfully mounted prefix 'prefix' of bucket 'bucket' at /data/s3/prefix
And then we get back to list the content of the S3 mountpoint to find it empty.
From the "mount-s3 --debug --debug-crt" we saw that the mountpoint is automatically unmounted for some reason:
2024-01-09T12:09:52.753567Z INFO fuser::session: Mounting- /data/s3/prefix
2024-01-09T12:09:52.754068Z INFO mount_s3: successfully mounted prefix 'prefix' of bucket 'bucket' at /data/s3/prefix
2024-01-09T12:09:52.754152Z DEBUG mount_s3: success status flag received from child process
2024-01-09T12:09:52.755034Z DEBUG mountpoint_s3::fuse::session: starting fuse worker 0 (thread id 9586)
2024-01-09T12:09:52.755055Z DEBUG fuser::request: FUSE( 1) ino 0x0000000000000000 INIT kernel ABI 7.26, capabilities 0x1ffffb, max readahead 131072
2024-01-09T12:09:52.755099Z DEBUG fuser::request: INIT response: ABI 7.28, flags 0x2021, max readahead 131072, max write 16777216
2024-01-09T12:09:52.784233Z INFO fuser::session: Unmounted /data/s3/prefix
2024-01-09T12:09:52.784265Z DEBUG awscrt::S3Client: id=0x560d6b3316f0 Client starting destruction.
2024-01-09T12:09:52.784318Z DEBUG awscrt::channel-bootstrap: id=0x560d6b31cc40: releasing bootstrap reference
Direct listing via "aws s3 ls" works fine.
Complete log is attached to "relevant log output" section.
Relevant log output
Will try to attach via a dedicated file
The text was updated successfully, but these errors were encountered:
Hey @sedrichaus, sorry you're seeing this issue. Those logs do not give us any information on what happened, do they? 😅
Let's try this instead - I trust this issue is frequently reproducible.
Do not use --debug and --debug-crt, and instead set the environment variable MOUNTPOINT_LOG=trace,awscrt=warn. This will configure Mountpoint's logging to put everything out at trace level, excluding the AWS CRT. I suspect that the CRT is not the issue here. There's more information on this particular logging configuration in Mountpoint's logging documentation.
Mountpoint for Amazon S3 version
1.3.1
AWS Region
us-east-1
Describe the running environment
Mountpoint options
mount-s3 bucket --prefix prefix /data/s3/prefix --max-threads=1 --debug --log-directory /var/log/mount-s3 Same behavior observed if executed with the required options only mount-s3 bucket /data/s3/prefix
What happened?
And then we get back to list the content of the S3 mountpoint to find it empty.
From the "mount-s3 --debug --debug-crt" we saw that the mountpoint is automatically unmounted for some reason:
Direct listing via "aws s3 ls" works fine.
Complete log is attached to "relevant log output" section.
Relevant log output
The text was updated successfully, but these errors were encountered: