FAQs

Common troubleshooting

1. When bosfs is used, it suddenly displays "unable to access MOUNTPOINT /path/to/mountpoint: Transport endpoint is not connected", and becomes inaccessible.

   • First, you can check whether the bosfs process exists using the ps ax | grep bosfs command. If the bosfs process is terminated due to an incorrect operation, you can remount it by executing the following commands:
   Plain Text Copy

   1        fusermount -uz /path/to/mnt_dir
   2        bosfs bucket /mnt/bosfs xxxx

   • If the bosfs process termination is unrelated to user error, verify whether the machine's fuse version is below 2.9.4. libfuse versions earlier than 2.9.4 may induce abnormal bosfs process termination. In such scenarios, it is advisable to update fuse or install the latest bosfs version as guided in this document's bosfs compilation and installation section.
   • bosfs occupies disk space for writing a large quantity temp caches during read/write operations (it is in the /tmp directory by default, and the file directory can be specified using -o tmpdir=/mnt/tmp-path), which may lead to insufficient available disk space and requires frequent cleanup.
   • The possible reason is that listbucket or listobject requests excessive memory, triggering the system OOM.
   • boscmd is recommended instead of bosfs only in data download/upload scenarios, bosfs can be used for non-sensitive online services, and bcecmd is recommended in scenarios requiring reliability and stability.

2. Prolonged freezing during usage.

   • Check whether it is caused by high disk I/O load. It is recommended that the concurrency should be reduced and -o multipart_parallel=5 should be specified during mounting (the default value is 10).
   • Check the network environment from the machine to the BOS endpoint. In a public network or unstable network scenario, you can configure -o client_timeout=5000 so that timeout occurs and retry is conducted after the configured duration of network transmission expires. The timeout duration can be set according to the specific network environment.
   • Enable logging to view other specific reasons.

3. Inconsistent file metadata when multiple bosfs processes mount the same bucket

   • Using meta cache may cause inconsistent metadata (such as length and mod permission) across machines. Specify-o meta_expires=0 during mounting to avoid reading metadata from cache, but this may increase duration of ls/stat and other commands.

4. Error "there is no enough disk space for used as cache(or temporary)" is displayed during large file upload

   • bosfs uses the /tmp directory to store temporary files during upload and download. If the disk space in this directory is insufficient, an error will be reported.
   • If the available disk space in the /tmp directory is less than the file size, increase the space in /tmp.
   • Check the configuration of multipart_size and multipart_parallel to ensure they are set correctly, and verify that the available space in the /tmp/ directory is greater than multipart_size * multipart_parallel.
   • For scenarios involving only uploads and downloads (sequential read/write), you can enable memory cache mode with the -o rw_strategy=mem_sequence -o mem_cache_per_size=20MB configuration, and the /tmp/ directory will no longer be used.

5. 403 error encountered during use

   • Generally, 403 errors are caused by permission-related issues; it is suggested that AK and permission configuration should be checked.