Hi, Tsunakawa-san, Kirk-san. Thank you for your review.
Tsunakawa-san, > From: Tsunakawa, Takayuki [mailto:tsunakawa.ta...@jp.fujitsu.com] > The client-side tcp_user_timeout patch looks good. Thanks, but I minorly changed that patch. It is the declaration place of sebuf[], moving to just before where it'll become needed. > The server-side tcp_user_timeout patch needs fixing the following: > (1) > + GUC_UNIT_MS | GUC_NOT_IN_SAMPLE > + 12000, 0, INT_MAX, > > GUC_NOT_IN_SAMPLE should be removed because the parameter appears in > postgresql.conf.sample. > The default value should be 0, not 12000. Oops, fixed! > (2) > Now that you've changed show_tcp_usertimeout() to use > pq_gettcpusertimeout(), you need to modify pq_settcpusertimeout(); just > immitate pq_setkeepalivesidle(). > Otherwise, SHOW would display a wrong value. Indeed, pq_settcpusertimeout() should be modified as well. > The socket_timeout patch needs the following fixes. Now that others > have already tested these patches successfully, they appear committable to me. Thank you so much for reviewing even socket_timeout patch. > The reason why oom_error label is present is that it is used at > multiple places to avoid repeating the same error processing code. Understood. > (2) > + conn->sock = -1; > > Use PGINVALID_SOCKET instead of -1. Oh, I see. Kirk-san, > I referred to the doc in libpq.sgml, corrected misspellings, and rephrased the > doc a little bit as below: Changed "stop-gap" and "mean" -> "stopgap" and "measure" respectively. But I can't find the difference between mine and yours. Would you please tell me if there is a difference like this depending on your rephrasing? Best regards, --------------------- Ryohei Nagaura
socket_timeout_v10.patch
Description: socket_timeout_v10.patch
TCP_backend_v17.patch
Description: TCP_backend_v17.patch
TCP_interface_v17.patch
Description: TCP_interface_v17.patch