Re: Are indexes blown?
От | Andrej Ricnik-Bay |
---|---|
Тема | Re: Are indexes blown? |
Дата | |
Msg-id | b35603930802161147x60e0d7bevd4b68a7b92227c67@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: Are indexes blown? (Greg Smith <gsmith@gregsmith.com>) |
Ответы |
Re: Are indexes blown?
|
Список | pgsql-general |
On 16/02/2008, Greg Smith <gsmith@gregsmith.com> wrote: > top -bc | tee topdata > > That will save everything to a file called topdata while also letting you > watch it scroll by. Not as easy to catch the bad periods that way, the > output is going to be a huge data file, but you'll have a log to sort > through of everything. Control-C to get out of there when you're bored. Or pipe through the following awk-script to make the output less overwhelming ... save it as top.awk BEGIN{ # the sort numerically sorts by memory usage percentage # head filters out the top 20 contenders command = "LC_ALL=C sort -k 10,10gr|head -20" } { # read all of top into an array line[NR]=$0 last=NR } END{ # print the header well-formed for(i=1;i<8;i++){ print line[i] } # and do the sort & strip of the processes for(i=8;i<last;i++){ print line[i]|& command } close(command, "to") while ((command |& getline out) > 0) print out close(command) } Invoke like so top -b -d 1 | awk -f top.awk | tee topdata Instead of the "getting bored Ctrl-C" maybe a "-n 3600" as extra parameter to top to get roughly one hours worth of data... Cheers, Andrej
В списке pgsql-general по дате отправления: