I have a Cassandra cluster of 6 VMs, each of these VMs has an NFS share attached, the last few weeks the backups from some nodes began to fall off, began debugging, as a result I tried to write zeros, that's what happened in the end:
1vm
time dd if = / dev / zero of = foel bs = 16k count = 128k
131 072 + 0 records in
131072 + 0 records out
2147483648 bytes (2.1 GB) copied, 4.23151 s, 507 MB / s
real 0m4.239s
user 0m0.008s
sys 0m1.593s
2vm
time dd if = / dev / zero of = foel bs = 16k count = 128k
131 072 + 0 records in
131072 + 0 records out
2147483648 bytes (2.1 GB) copied, 4.83356 s, 444 MB / s
real 0m4.839s
user 0m0.005s
sys 0m1.544s
3vm
time dd if = / dev / zero of = foel bs = 16k count = 128k
131072 + 0 records in
131072 + 0 records out
2147483648 bytes (2.1 GB) copied, 56.0399 s, 38.3 MB / s
real 0m56.054s
user 0m0.138s
sys 0m1.741s
4vm
time dd if = / dev / zero of = foel bs = 16k count = 128k
131072 + 0 records in
131072 + 0 records out
2147483648 bytes (2.1 GB) copied, 6.88073 s, 312 MB / s
real 0m6.887s
user 0m0.125s
sys 0m1.862s
5vm
time dd if = / dev / zero of = foel bs = 16k count = 128k
131 072 + 0 records in
131072 + 0 records out
2147483648 bytes (2.1 GB) copied, 48.762 s, 44.0 MB / s
real 0m48.771s
user 0m0.114s
sys 0m1.740s
6vm
time dd if = / dev / zero of = foel bs = 16k count = 128k
131072 + 0 records in
131072 + 0 records out
2147483648 bytes (2.1 GB) copied, 5.74801 s, 374 MB / s
real 0m5.754s
user 0m0.104s
sys 0m1.744s
As you can see, at 5 and 3 vm, writing is very slow.
Mount options are the same on all VMs:
tac / etc / fstab
172.24.5.229:/kosandra_bekap / opt / cassandra-backup nfs rw, noatime 0 0
I can't ping the nfs, pings are closed.