HTTP -- 502 Bad Gateway

Dear ORS,

Recently I’ve experienced a lot of 502 bad gateway errors around the Matrix API. It looks to me as if a reply from a proxy is timing out, as the 502 happens consistently at the 1 minute mark (1.09s).

So far I’ve seen that using a smaller matrix size 6x6 is working as it takes less time to resolve.
Am I doing something wrong?

A sample of my original query:

{
“locations”: [[4.938574,52.404254],[6.416581,52.530271],[4.392741,52.057036],[5.088932,51.739693],[6.572343,53.008928],[4.871871,51.660023],[4.780813,52.155521],[5.18499,51.586974],[5.085441,51.577416],[5.779177,51.742695],[4.861313,51.947939],[6.660562,52.351922],[5.555264,51.615665],[5.693136,51.458575],[6.759829,52.243913],[4.950915,51.843195],[4.466534,52.134268],[4.291974,51.53868],[5.938097,51.635522],[5.844193,50.99651],[4.506073,52.048609],[5.316014,51.864191],[6.088112,53.109017],[4.769521,51.60141],[4.440487,52.209215],[6.504089,53.335411],[4.3094,51.849874],[5.069207,52.499906],[6.762564,52.244023],[5.939294,52.933121],[5.847102,51.833609],[5.073786,51.682315],[5.067633,51.589288],[6.377426,51.891431],[4.997931,52.518493],[6.27141,51.973671]],
“metrics”: [“distance”, “duration”],
“units”: “m”
}

I am also experiencing the same issue.

It has been days now.

Hopefully it’s going to be addressed soon.

I think we found the issue, now we just have to hot-fix it. We’ll update the open threads once it’s fixed.

Thank you for the reply Nils! That might have just saved my thesis :stuck_out_tongue:

Cheers!

Oops;) Well, if you need it only for a smallish region like Baden-Württemberg/half a country, you can always quickly install it yourself:

Interestingly it’s all fine that way, at least for driving-* profiles. foot-walking is somehow super slow, even locally.

fixed nowwwwwwwwwwww

Cheers Nils, You guys are amazing!

1 Like

So we got some 502s repeatedly for one set of points today. About 1:45PM Oslo time today if that helps anything. :slight_smile:

It seems that the car profile has been under heavy load from matrix and isochrone requests lately, so when too many are asked for at the same time, they basically hog all of Tomcat’s threads until the tasks are complete, thus giving the timeout. When multiple heavy requests are run one after the other, then it basically blocks the system indefinitely. I’ve just restarted the car server now so hopefully if some threads got stuck somewhere that will will release them