akosiaris added a comment.

  In T199219#4452041 <https://phabricator.wikimedia.org/T199219#4452041>, 
@Smalyshev wrote:
  
  > @BBlack I am getting rather strange result with 
`appservers-ro.discovery.wmnet` - if I call the URL you provided, the call 
takes a lot of time:
  >
  > real    0m4.270s
  >
  > while if I call to `www.wikidata.org`, I get:
  >
  > real    0m0.127s
  >
  > Same with `api-ro`. `appservers-rw` is a bit faster:
  >
  > real    0m0.320s
  >
  > But still 3x from going through frontend (and it's not caching - I changed 
the URL, result is the same, and varnish settings all say "miss").
  
  
  Is this still true? I see
  
    deploy1001:~$ for i in appservers-ro appservers-rw api-ro api-rw ; do echo 
-n $i; time curl -s -o /dev/null -X GET -H 'Host: www.wikidata.org' 
"https://${i}.discovery.wmnet/wiki/Special:EntityData/Q2408871.ttl?nocache=1530836328152&flavor=dump";
 ; done
    appservers-ro
    real        0m0.097s
    user        0m0.020s
    sys 0m0.012s
    appservers-rw
    real        0m0.113s
    user        0m0.028s
    sys 0m0.000s
    api-ro
    real        0m0.113s
    user        0m0.024s
    sys 0m0.012s
    api-rw
    real        0m0.128s
    user        0m0.028s
    sys 0m0.004s
  
  Note that it's quite important where the tests are run from. That is the 
active DC is going to be faster anyway. Running those from codfw yields 
entirely different results as there are back and forths between the DCs that 
need to be served.
  
    deploy2001:~$ for i in appservers-ro appservers-rw api-ro api-rw ; do echo 
-n $i; time curl -s -o /dev/null -H 'Host: www.wikidata.org' 
"https://${i}.discovery.wmnet/wiki/Special:EntityData/Q2408871.ttl?nocache=1530836328152&flavor=dump";
 ; done
    appservers-ro
    real        0m6.435s
    user        0m0.024s
    sys 0m0.012s
    appservers-rw
    real        0m0.323s
    user        0m0.012s
    sys 0m0.020s
    api-ro
    real        0m5.061s
    user        0m0.028s
    sys 0m0.004s
    api-rw
    real        0m0.276s
    user        0m0.032s
    sys 0m0.000s
  
  Which are comparable to the numbers posted above.

TASK DETAIL
  https://phabricator.wikimedia.org/T199219

EMAIL PREFERENCES
  https://phabricator.wikimedia.org/settings/panel/emailpreferences/

To: akosiaris
Cc: akosiaris, BBlack, Aklapper, Smalyshev, Gehel, darthmon_wmde, Premeditated, 
Nandana, Lahi, Gq86, Lucas_Werkmeister_WMDE, GoranSMilovanovic, QZanden, 
EBjune, merbst, LawExplorer, _jensen, rosalieper, Jonas, Xmlizer, jkroll, 
Wikidata-bugs, Jdouglas, aude, Tobias1984, Manybubbles, Mbch331
_______________________________________________
Wikidata-bugs mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikidata-bugs

Reply via email to