[Linux] Zip operations are extremely slow in juicefs directory

[0mza987]

What happened:
I use below command to mount juicefs:

juicefs mount --subdir ${JUICEFS_SUBDIR} redis://${user}:${password}@${endpoint}/${redis_db} ${mount_path} --cache-dir /var/jfsCache/ -o allow_other,nonempty \
  --writeback --attr-cache=1 --free-space-ratio=0.2 --entry-cache=1 --dir-entry-cache=1 --negative-entry-cache=1 --readdir-cache=true \
  --open-cache=1 --prefix-internal ${metrics_flag} --log="/var/log/juicefs.log" &

The latency is OK when read/write files, but i found it extremely slow when zipping files with zip -q -r xxx, is this by design or i have some config wrong in my mount command?

Thanks

What you expected to happen:
zip operations have a normal speed.

How to reproduce it (as minimally and precisely as possible):
use zip -q -r to compress folders in juicefs folder.

Anything else we need to know?

Environment:

• JuiceFS version (use juicefs --version) or Hadoop Java SDK version: 1.3.0+2025-07-03.30190ca
• Cloud provider or hardware configuration running JuiceFS: Alibaba Cloud
• OS (e.g cat /etc/os-release): alinux3
• Kernel (e.g. uname -a):
• Object storage (cloud provider and region, or self maintained): Alibaba OSS
• Metadata engine info (version, cloud provider managed or self maintained):
• Network connectivity (JuiceFS to metadata engine, JuiceFS to object storage):
• Others:

[jiefenghuang]

zip involves a lot of metadata operations, such as lookup, getattr, etc. The above configuration is fine, but you need to increase the time settings in order to effectively reduce metadata access — adjust them according to your needs.

You can capture the operation log during your zip run with:

cat /mount_point/.accesslog > tmp.log

Then analyze it with:

juicefs profile tmp.log --interval 0

to see what operations are being performed most frequently, and then adjust your cache strategy accordingly.