Lengthy runs of algorithms hit a timeout

TigerGraph DB
#1

Hi,

I’ve installed a pretty large (120M edges, ~4bn vertices) graph. When applying pagerank on it, I get

408 Client Error: Timeout for url: http://127.0.0.1:9010/query/mygraph/tg_pagerank

Im running this via pyhton:


conn = tg.TigerGraphConnection(
    host="http://127.0.0.1",
    graphname="mygraph",
    username="tigergraph",
    password="XXXXX",
    restppPort=9010)

f=conn.gds.featurizer()

params = {'v_type':'Person','e_type':'KNOWS',
     'max_change':0.001, 'max_iter': 25, 'damping': 0.85, 'top_k': 100,
     'print_accum': False, 'result_attr':'pagerank','file_path':'','display_edges': False}
f.runAlgorithm('tg_pagerank', params=params)

Config option RESTPP.Factory.DefaultQueryTimeoutSec is already set way larger: 19200.

Any ideas what the cause of that timeout could be?

#2

Hmmm. I think PageRank should be returning within the default timeout that we set in the featuirizer, but you can try changing the f.runAglorithm() call to include the timeout parameter, so it would be:

f.runAlgorithm('tg_pagerank', params=params, timeout=19200000).

Hopefully this works for you, and I will followup with the expected execution time of PageRank on a graph of that size.

1 Like
#3

If you could also provide any details about your environment (RAM, CPU, etc.) and what version of TigerGraph you are using would be great!

#4

Thanks a lot, adding timeout parameter did the trick.

2 Likes
#5

@grizard Thanks for replying back to the tread and providing the resolution!