rorton Posted October 20, 2018 Share Posted October 20, 2018 I made a post before, summary is that I've had a Crucial MX500 installed as a cache drive for just under 3 months, and the SMART data is saying that i have used 17% of the SSD's life, and i have 83% remaining. Specs of the drive suggest an endurance of 100TB Total Bytes Written (TBW), equal to 54GB per day for 5 years. Considering i don't write a massive amount to the device anyway, this should be fine. On my SSD as cache, i have my docker image (21gb in size), with things like SAB, Sonaar, Radaar, Emby installed, and i have a VM with Linux running the SNMP app LibreNMS. I have done an iotop on the machine, and have had it running for 4 hours in cumulative mode, and have the below output: Total DISK READ : 0.00 B/s | Total DISK WRITE : 148.68 K/s Actual DISK READ: 0.00 B/s | Actual DISK WRITE: 141.92 K/s PID PRIO USER DISK READ DISK WRITE> SWAPIN IO COMMAND 5479 be/4 root 128.16 M 2.52 G 0.00 % 0.16 % qemu-system-x86_64 -name guest=LibreNMS,debug-threads=on -S -object secret,id=masterKey0,format=raw,file=/var/lib/libvirt/qemu/domain-1-LibreNMS/m~rtio-balloon-pci,id=balloon0,bus=pci.4,addr=0x0 -sandbox on,obsolete=deny,elevateprivileges=deny,spawn=deny,resourcecontrol=deny -msg timestamp=on 8623 be/4 ortonr 172.58 M 1382.31 M 0.00 % 5.13 % afpd -d -F /etc/netatalk/afp.conf 5741 be/4 nobody 73.76 M 1376.48 M 0.00 % 0.01 % bin/mongod --dbpath /usr/lib/unifi/data/db --port 27117 --unixSocketPrefix /usr/lib/unifi/run --logappend --logpath /usr/lib/unifi/logs/mongod.log --bind_ip 127.0.0.1 4430 be/0 root 85.92 M 794.06 M 0.00 % 0.05 % [loop2] 4451 be/4 root 0.00 B 571.91 M 0.00 % 0.06 % [btrfs-transacti] 4318 be/4 root 168.00 K 435.67 M 0.00 % 0.03 % [btrfs-transacti] 4222 be/4 nobody 9.77 M 194.32 M 0.00 % 0.01 % mono --debug NzbDrone.exe -nobrowser -data=/config 3595 be/4 nobody 4.79 M 116.98 M 0.00 % 0.01 % mono --debug Radarr.exe -nobrowser -data=/config 8624 be/4 nobody 292.00 K 95.03 M 0.00 % 0.00 % cnid_dbd -F /etc/netatalk/afp.conf -p /mnt/disk3 -t 6 -l 4 -u ortonr 4607 be/4 nobody 6.50 M 16.88 M 0.00 % 0.00 % java -Xmx1024M -jar /usr/lib/unifi/lib/ace.jar start 5520 be/4 daemon 1360.92 M 5.88 M 0.00 % 0.00 % EmbyServer -programdata /config -ffmpeg /bin/ffmpeg -ffprobe /bin/ffprobe -restartexitcode 3 15808 be/4 root 8.00 K 5.08 M 0.00 % 0.16 % [kworker/u8:2-edac-poller] 25610 be/4 root 49.00 K 3.07 M 0.00 % 0.01 % [kworker/u8:0-btrfs-endio-write] 2306 be/4 root 0.00 B 1168.00 K 0.00 % 0.01 % [kworker/u8:3-btrfs-endio-write] 4464 be/4 root 0.00 B 940.00 K 0.00 % 0.00 % dockerd -p /var/run/dockerd.pid --storage-driver=btrfs --storage-driver=btrfs 7201 be/4 root 0.00 B 640.00 K 0.00 % 0.01 % [kworker/u8:1-btrfs-endio-write] 4317 be/4 root 0.00 B 192.00 K 0.00 % 0.00 % [btrfs-cleaner] 4347 be/4 root 1349.72 M 8.00 K 0.00 % 0.03 % shfs /mnt/user -disks 7 2048000000 -o noatime,big_writes,allow_other -o remember=0 4145 be/4 root 0.00 B 0.00 B 0.00 % 0.01 % emhttpd 4206 be/4 root 0.00 B 0.00 B 0.00 % 0.06 % [unraidd] 4285 be/4 root 2.75 M 0.00 B 0.00 % 0.00 % [xfsaild/md3] the biggest process writing data seems to be LibreNMS, and the VM i have running, which it recons has written 2.52gb in 4 hours, so just under a gig an hour. Even if that was a gig an hour, that's only 24gb per day, the other seems to be Unifi Docker, which has done 1gb in 4 hours, again, not a lot. i can't understand why the drive is reducing its lifespan so quickly. With the above writes, it should last the 5 years that say it will last, but based on the smart data, its only going to last 1 year. Plus, if LibreNMS is writing 2.52gb in 4 hours, (so say 20gb a day), where is all that data? The SSD is only 250gb, the SSD would be full in 10 days if this amount of data was being written? Cant get my head around it at all Link to comment
Frank1940 Posted October 20, 2018 Share Posted October 20, 2018 Is this SSD connected to the MB SATA port or to a added SATA expansion card? I seem to remember that a few SATA cards don't support the TRIM operation which is necessary to prolong SSD life and performance. Link to comment
rorton Posted October 20, 2018 Author Share Posted October 20, 2018 Thanks for replying. It’s connected to the motherboard. The machine is one of those small HP N40L microservers. Link to comment
rorton Posted October 20, 2018 Author Share Posted October 20, 2018 Forgot to mention. The trim plugin is installed and scheduled to run every day Link to comment
JorgeB Posted October 20, 2018 Share Posted October 20, 2018 I wouldn't worry too much for now, first the estimated life is just that, an estimation, people are known to continue using SSDs for years after SMART reports life at zero, if 17% represents around 17TB written, and it should be much less due to wright amplification, it's not that much for 3 months, my cache device is less than 2 years old, I don't use it to cache any shares, just a couple of dockers and 3 VMs, and this is the current writes total: Data Units Written: 225,064,266 [115 TB] Link to comment
rorton Posted October 21, 2018 Author Share Posted October 21, 2018 thanks - where does all the data go then, i dont get that in 12 hours, my LibreNMS VM has written 14.55 GB worth of data according to this iota app, and yet the available space on the SSD hasn't reduced by 14GB, Unifi has apparently written 7gb etc, Link to comment
JorgeB Posted October 21, 2018 Share Posted October 21, 2018 6 minutes ago, rorton said: thanks - where does all the data go then 38 minutes ago, rorton said: my LibreNMS VM has written 14.55 GB worth of data according to this iota app, and yet the available space on the SSD hasn't reduced by 14GB, Unifi has apparently written 7gb etc, It's normal not to reduce available space, it's overwriting, it's like for example a windows bare metal install, it keeps doing small writes to disk but it's not always using more space. In my case my VMs and dockers write about 12 to 15GB/hour Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.