Mbch331 added a comment.
When I add a LIMIT 1 to the query, I get a time-out.
TASK DETAIL
https://phabricator.wikimedia.org/T127014
EMAIL PREFERENCES
https://phabricator.wikimedia.org/settings/panel/emailpreferences/
To: Mbch331
Cc: Mbch331, Magnus, JanZerebecki, Smalyshev, Aklapper, Stu
Smalyshev added a comment.
See: https://jira.blazegraph.com/browse/BLZG-1815
TASK DETAIL
https://phabricator.wikimedia.org/T127014
EMAIL PREFERENCES
https://phabricator.wikimedia.org/settings/panel/emailpreferences/
To: Smalyshev
Cc: Mbch331, Magnus, JanZerebecki, Smalyshev, Aklapper, St
Nikki added a comment.
I'm having this problem with various queries too.
It seems to happen a lot with sitelink queries, e.g.
https://query.wikidata.org/#SELECT%20%3Fitem%20%3Farticle%0AWHERE%0A%7B%0A%09%3Farticle%20schema%3Aabout%20%3Fitem%20.%0A%09FILTER%20(SUBSTR(str(%3Farticle)%2C%201
Gehel added a comment.
Doing a tcpdump capture, it seems that both request directly to nginx on
wdqs1001 and requests going through LVS + varnish (see the curl in above
comment) arrive correctly at Blazegraph. Blazegraph sends response in both
cases. Taking tcpdumps in front of nginx or betw
Gehel added a comment.
Continuing investigation:
1. capture tcpdump from cp1045: `sudo tcpdump -i any 'host
wdqs1001.eqiad.wmnet or port 80' -w varnish.pcap`. This should capture both
ingress and egress traffic to varnish.
2. send query directly to varnish from cp1045: `curl -k -D -
'
Smalyshev added a comment.
@BBlack 4s looks way to low, especially given that Blazegraph does stream
results, so it may definitely happen than it has pauses between bytes. But what
looks weird to me is that empty request seems to finish even faster than 4s. In
any case, would it be hard to r
gerritbot added a comment.
Change 279157 had a related patch set uploaded (by Gehel):
Increasing between_bytes_timeout for wikidata query service
https://gerrit.wikimedia.org/r/279157
TASK DETAIL
https://phabricator.wikimedia.org/T127014
EMAIL PREFERENCES
https://phabricator.wikime
gerritbot added a comment.
Change 279157 merged by BBlack:
Increasing between_bytes_timeout for wikidata query service
https://gerrit.wikimedia.org/r/279157
TASK DETAIL
https://phabricator.wikimedia.org/T127014
EMAIL PREFERENCES
https://phabricator.wikimedia.org/settings/panel/emai
Smalyshev added a comment.
Looks like the problem still happens, and it takes less than 4s to happen,
namely time returns:
real 0m3.289s
Maybe it is some kind of data size limit?
TASK DETAIL
https://phabricator.wikimedia.org/T127014
EMAIL PREFERENCES
https://phabricator.wikime
BBlack added a comment.
I did some live experimentation with manual edits to the VCL. It is the
`between_bytes_timeout`, but the situation is complex. The timeout that's
failing is on the varnish frontend fetching from the varnish backend. These
are fixed at 2s, but because this is all in
Gehel added a comment.
Placing this task up for grabs as it is unclear to me what will need to be
done once https://phabricator.wikimedia.org/T128813 is resolved.
TASK DETAIL
https://phabricator.wikimedia.org/T127014
EMAIL PREFERENCES
https://phabricator.wikimedia.org/settings/panel/emai
Smalyshev added a comment.
Confirmed, the query runs fine now, thanks @BBlack!
TASK DETAIL
https://phabricator.wikimedia.org/T127014
EMAIL PREFERENCES
https://phabricator.wikimedia.org/settings/panel/emailpreferences/
To: BBlack, Smalyshev
Cc: gerritbot, BBlack, Gehel, Nikki, Mbch331, Ma
12 matches
Mail list logo