Score:0

sshfs slow from one machine but not from another

cn flag

Situation

I have two web servers that connect to a file repository over sshfs:

/usr/bin/sshfs -f -o reconnect -o follow_symlinks -o allow_other \
  repo@files3:/var/repo /var/repo

On one web server this mount point can be used with "normal" speed, but on the other web server any operation on the mounted folder takes a long time, for example:

user@good-server$ time ls /var/repo/big-folder > /dev/null

real    0m0.513s
user    0m0.091s
sys 0m0.048s

vs

user@bad-server$ time ls /var/repo/big-folder > /dev/null

real    0m15.414s
user    0m0.049s
sys 0m0.006s

I can reproduce this slowness with plain ssh (e.g. time ( for i in {1..10}; do ssh repo@files3 true; done) takes 1.5 vs 7 seconds).

Dead ends

I have a mirrored backup server of the files3 (called files4) for which I can reproduce the slowness with the bad webserver and the speed with the good webserver. So I conclude that the problem is on the ssh client not the server. The /etc/ssh/sshd_config is the same on all four servers though. All the machines are also listed in each others /etc/hosts files with ipv4 and ipv6 address (mentioning this due to SSH connection slow from one machine but not another and Improving SSHFS performance when reflecting changes in host directory).

Question

What could be the cause or how can I go about debugging this?

I sit in a Tesla and translated this thread with Ai:

mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.