Hi James,

I am not sure whether i have to add more RAM or change the disk we are
using for caching. I am using normal SATA (7k rpm) as raw disk for caching.

Should I change it to SAS or SSD?  Anyone is using RAMDISK for caching as
disk?

Or I have to add more RAM?

Currently there is only one disk I am using for caching. Will it help if I
increase the number of raw disks for caching?

Regards,
Randeep



On Wed, Jul 13, 2016 at 4:30 AM, James Peach <jpe...@apache.org> wrote:

>
> > On Jul 12, 2016, at 6:05 AM, Randeep <randeep...@gmail.com> wrote:
> >
> > Hi all,
> >
> > I'm trying to understand what is causing the load on the ats server. Not
> sure whether is RAM or disk or CPU is giving the issues.
>
> From the output you posted it seems pretty likely that you are IO bound on
> reads. You should investigate why that is happening. For most caching
> workloads, I would expect the RAM cache to give a good hit rate, so that is
> one avenue of investigation.
>
> > on top output I can see a lot of iowait.
> >
> > top - 18:12:00 up  4:32,  3 users,  load average: 8.15, 8.20, 8.53
> > Tasks: 234 total,   2 running, 232 sleeping,   0 stopped,   0 zombie
> > Cpu(s):  7.2%us, 10.7%sy,  0.0%ni, 34.0%id, 46.5%wa,  0.0%hi,  1.7%si,
> 0.0%st
> > Mem:   7848208k total,  7701228k used,   146980k free,    20008k buffers
> > Swap:  8388604k total,   426796k used,  7961808k free,   250532k cached
> >
> >   PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
> > 14138 ats       20   0 8199m 6.7g 4852 S  9.0 89.8   7:23.41
> /opt/ats/bin/traffic_server -M --bind_stdout
> /opt/ats/var/log/trafficserver/traffic.out --bind_stderr /opt/
> >
> > And on iostat,
> > Device:         rrqm/s   wrqm/s     r/s     w/s   rsec/s   wsec/s
> avgrq-sz avgqu-sz   await  svctm  %util
> > sda               0.00    83.00  257.00   54.00 134672.00 21248.00
>  501.35    10.51   33.69   3.22 100.00
> >
> > on iotop:
> > Total DISK READ: 50.46 M/s | Total DISK WRITE: 14.35 M/s
> >   TID  PRIO  USER     DISK READ  DISK WRITE  SWAPIN     IO>    COMMAND
> > 14154 be/4 ats         3.33 M/s    0.00 B/s  0.00 % 91.45 %
> traffic_server -M --bind_stdout
> /opt/ats/var/log/traf~s/var/log/trafficserver/traffic.out --httpport 80:fd=9
> > 14156 be/4 ats         7.28 M/s 2025.05 K/s  0.00 % 90.56 %
> traffic_server -M --bind_stdout
> /opt/ats/var/log/traf~s/var/log/trafficserver/traffic.out --httpport 80:fd=9
> > 14157 be/4 ats         6.88 M/s    0.00 B/s  0.00 % 90.52 %
> traffic_server -M --bind_stdout
> /opt/ats/var/log/traf~s/var/log/trafficserver/traffic.out --httpport 80:fd=9
> > 14155 be/4 ats         2.53 M/s    3.32 M/s  0.00 % 82.89 %
> traffic_server -M --bind_stdout
> /opt/ats/var/log/traf~s/var/log/trafficserver/traffic.out --httpport 80:fd=9
> > 14158 be/4 ats         9.58 M/s    3.33 M/s  0.00 % 75.91 %
> traffic_server -M --bind_stdout
> /opt/ats/var/log/traf~s/var/log/trafficserver/traffic.out --httpport 80:fd=9
> > 14161 be/4 ats         7.67 M/s    0.00 B/s  0.00 % 73.36 %
> traffic_server -M --bind_stdout
> /opt/ats/var/log/traf~s/var/log/trafficserver/traffic.out --httpport 80:fd=9
> > 14159 be/4 ats         5.89 M/s 2025.05 K/s  0.00 % 73.27 %
> traffic_server -M --bind_stdout
> /opt/ats/var/log/traf~s/var/log/trafficserver/traffic.out --httpport 80:fd=9
> > 14160 be/4 ats         7.30 M/s    3.75 M/s  0.00 % 66.01 %
> traffic_server -M --bind_stdout
> /opt/ats/var/log/traf~s/var/log/trafficserver/traffic.out --httpport 80:fd=9
> >
> > But I have disabled all the logs. I don't know why it still shows. On
> production, usually everyone disables the logs?
> >
> > [root@main ~]# free -m
> >              total       used       free     shared    buffers     cached
> > Mem:          7664       7488        175          0         13        210
> > -/+ buffers/cache:       7265        398
> > Swap:         8191        416       7775
> > [root@main ~]# w
> >  18:15:25 up  4:35,  3 users,  load average: 8.07, 8.14, 8.43
> >
> > I am using ats to cache live channels. Currently I have enabled 45
> channels. We have to enable 400 channels. So I am checking what resources I
> have to add.
> >
> > Any suggestions will be appreciated
> >
> > Best Regards,
> > --
> > Randeep
> > Mob: +919447831699[kerala]
> > Mob: +919880050349[B'lore]
> > http://twitter.com/Randeeppr
> > http://in.linkedin.com/in/randeeppr
> >
> >
> >
> > Randeep Raman
> > about.me/Randeeppr
> >
> >
>
>


-- 
Randeep
Mob: +919447831699[kerala]
Mob: +919880050349[B'lore]
http://twitter.com/Randeeppr
http://in.linkedin.com/in/randeeppr


[image: --]
Randeep Raman
[image: http://]about.me/Randeeppr
<http://about.me/Randeeppr>

Reply via email to