Enormous disk writes since 12.2015


(Thomas Eichstädt-Engelen) #1

Hi,

we are running the discourse docker installation on a digitalOcean 1GB instance. Everything way fine but something changes on December (around) 20th. From then on the disk writes (and hence the CPU consumptions) increased tremendously (see graph).

Could that be explained with changes in the Software? We always applies updates as they became available but unfortunately i didn’t find any history which updates have been applied when. What kind of logging, meta, support-data would you need to help nailing this problem down?

What i’ve double checked already:

  • normal amount of new users (20) per day
  • normal amount of posts (100-300) per day

Next i will ask digitialOcean if that behaviour could also be explained with different/broken hardware.

Appreciate any help, Thomas E.-E.


#2

Do you have any plugins or add anything in the customize section ?

I have pretty much the same stats and I have the same droplet, and I don’t have that problem.

Also, I reboot my server everyday, I think it could help. Connect in your droplet with putty and type this code : shutdown -r now
It will reboot your server


(Richard - DiscourseHosting.com) #3

You shouldn’t have to do that…


(Thomas Eichstädt-Engelen) #4

Plugins are:

  • discourse-details (0.3)
  • discourse-tagging (0.2)
  • docker_manager (0.1)
  • lazyYT (1.0.1)
  • poll (0.9)

and we added a new color theme to the section “customzize”


(Sam Saffron) #5

What does:

free -h
df -h

return?

Are you running out of memory?


(Thomas Eichstädt-Engelen) #6
             total       used       free     shared    buffers     cached
Mem:          994M       934M        59M        33M       2.7M        57M
-/+ buffers/cache:       874M       119M
Swap:         2.0G       688M       1.3G

and

Filesystem      Size  Used Avail Use% Mounted on
/dev/vda1        30G   16G   13G  55% /
none            4.0K     0  4.0K   0% /sys/fs/cgroup
udev            487M  4.0K  487M   1% /dev
tmpfs           100M  336K  100M   1% /run
none            5.0M     0  5.0M   0% /run/lock
none            498M  772K  497M   1% /run/shm
none            100M     0  100M   0% /run/user

restart of the whole machine has been issued two hours ago.


(Sam Saffron) #7

Looks to me like you are a bit tight on memory. If you get into a state where it is swapping, expect massive disk usage.


(Jeff Atwood) #8

Yes, consider this a sign that you should upgrade to a 2gb ram droplet.


(Thomas Eichstädt-Engelen) #9

ok, sounds reasonable … but why did that change on one day in December? By chance the straw that broke the camel’s back?


(Joshua Rosenfeld) #10

I don’t have any details for you on what would have caused the change, but your chart (to me) does not look like it changed in one day. To me it looks like Disk is increasing at a fairly steady (but low) rate since July, and once your usage exceeded the available RAM (around mid-December) the Disk writes increased at a much greater rate as you had no RAM to help lower the swap load.


(Thomas Eichstädt-Engelen) #11

you could be right … i’ve increased the RAM to 2GB. Let’s keep our fingers crossed it helps :slight_smile: