[ic] How to determine cause of load spikes.

[Kevin Walsh]

DB <DB at m-and-d.com> wrote:
> I'm running IC 5.4.1 using a mysql database of about 500,000 items on a
> dual Xeon box with 4GB of RAM. Once in awhile I'll notice the site
> become sluggish. During these periods the cpu load is always fairly
> high.  I suspect these events are caused by an inefficient search on the
> large products database. See below for an example 'top' output during a
> recent event:
> 
> top - 11:26:03 up 31 days, 20:50,  1 user,  load average: 1.13, 0.97, 0.67
> Tasks: 227 total,   2 running, 223 sleeping,   0 stopped,   2 zombie
> Cpu(s): 25.1% us,  0.1% sy,  0.0% ni, 74.9% id,  0.0% wa,  0.0% hi,  0.0% si
> Mem:   4040624k total,  3568748k used,   471876k free,   118736k buffers
> Swap:  2031608k total,     8664k used,  2022944k free,  1453916k cached
> 
>   PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
> 26248 inter     25   0 1367m 1.2g 1880 R  100 32.4   0:47.43 interchange
> 
> So I can identify the offending process ID, but how can I determine what
> this process is doing to cause such a load? If I can determine what
> search is being run or which of my pages is being accessed then I can
> probably correct the problem.
> 
> Bumping the RAM up to 4GB has drastically reduced the extent of the
> problem, but I want to find and correct the real cause of the trouble.
> 
I imagine, from the memory usage, that the cause might be an inefficiently
crafted Interchange search/scan operation.  The search/scan might be
forcing all, or a large chunk, of your products table into memory.  Check
the indexes on your table(s), check the search/scan itself and consider
using SQL for the search - if you're not already doing so.

-- 
   _/   _/  _/_/_/_/  _/    _/  _/_/_/  _/    _/
  _/_/_/   _/_/      _/    _/    _/    _/_/  _/   K e v i n   W a l s h
 _/ _/    _/          _/ _/     _/    _/  _/_/    kevin at cursor.biz
_/   _/  _/_/_/_/      _/    _/_/_/  _/    _/