M
M
Max2018-09-25 23:01:35
linux
Max, 2018-09-25 23:01:35

The new HDD goes into read-only under the load of sites, what should I do?

Good afternoon!
40 days ago I took a server with a Ryzen processor and 2x500 SSD ( sda+sdb ) + 1x4TB HDD ( sdc ) disks, the server hung up every 12 hours. After correspondence with the support, it turned out that the problem was in the HDD and the disk was replaced with a new one (as they said).
40 days have passed, a slight load has appeared on the server, and for the 2nd day in a row the web server stops working after the HDD on which the site files are located (the databases, and the system itself are on SSD). All this coincided with an almost minimal load, before that it was almost completely absent.
On top of that, when trying to write the cache file to disk, Apache creates up to 100 processes that hang, and as a result, the web server stops processing requests altogether, Apache processes just hang. I decided to temporarily reboot Apache every 5 minutes.
-Iqps54XT9uRcnOr_yaFKQ.pngRzLo9k9CRkWFHVsjm_CY5g.pngl5E1CQj5T-qaNUGX4sstNQ.png
Yesterday I did fsck -f -y /dev/sdc
The check went pretty quickly, after which I rebooted the server completely and the disk became RW again.
Here is the data /var/log/syslog
pastebin.com/PtFjWpzk
Today I did the same, the check took longer and there were more errors, I decided to run SMART - I'm waiting for the check to work, here's the screen before checking the current one. 5baa980234961512783907.jpeg
After half an hour of server operation after the reboot, 96 Bad Sectors have already appeared, before that there were 200 ... :
bDkbrSK8SHCtuNyS-q-zCQ.png
Content on the disk - 2 TB out of 4 TB, 1 TB - archives and dumps, 1 TB - files are mostly small (html page cache files, pictures). I checked iostat - the utilization was at the level of 5-10%, no more, this is already when the disk is only in Read Only.
Q9vXUAlrSxuyfEo-ie43DA.png
This is what the load looks like after a reboot and the disk is in RW, before the errors again rained down.
TIAloaXXS5yCWTvLpHI_ag.png
Questions:
1. Why can a disk go to Read-only and is it related to the load?
2. Can a disk become unusable in a month and what are the "bad tips" for this?
3. What to do next?
syslog log / dmesg pastebin.com/PtFjWpzk
fsck log pastebin.com/eSqeuFJc

Answer the question

In order to leave comments, you need to log in

3 answer(s)
P
Puma Thailand, 2018-09-27
@opium

Apparently the disk is dying again, if it was good at the beginning, then there is a possibility that your problems with the disk are connected either with the cable or with the controller or with power

J
jcmvbkbc, 2018-09-25
@jcmvbkbc

# fsck -f -y /dev/sdc
...
Warning! /dev/sdc is mounted.
...
You do not need to check the mounted disk, it will not get better from this.
If you feel like poking around, unmount it first.

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question