From mboxrd@z Thu Jan 1 00:00:00 1970 From: Daniel.Thibault@drdc-rddc.gc.ca (Thibault, Daniel) Date: Fri, 31 May 2013 19:48:15 +0000 Subject: [lttng-dev] Interplay of enable-channel options Message-ID: <48CF5AC71E61DB46B70D0F388054EFFD12D2256E@VAL-E-02.valcartier.drdc-rddc.gc.ca> Harking back to Mathieu's post of 3 May 2013 10:42:22, I'd like to see if I understand the options correctly. I'm pretty sure I don't. When a channel has been set to use per-UID buffering, if an event-producing user-space process dies suddenly it may create a "hole" in one of the channel's sub-buffers, temporarily preventing the consumer from getting access to the sub-buffer. Once the other event producers have filled the remaining sub-buffers, the session daemon will try to re-use the "hung" sub-buffer, at which point it will be discarded. Unrelated events in the sub-buffer will be lost. If the channel is also set to --discard, does this mean the session daemon will not flush the hung sub-buffer? This would mean the consumer daemon waits at the holed sub-buffer and never gets past it. What if --read-timer has been set to a non-zero value? Will the holed sub-buffer be flushed at that point in time? In this case the producers may not even have time to fill the remaining sub-buffers. How can the session daemon create a hole in a sub-buffer, anyway? Doesn't it get all of the event's description and payload in its own working memory before starting to write it out to the sub-buffer? Daniel U. Thibault Protection des syst?mes et contremesures (PSC) | Systems Protection & Countermeasures (SPC) Cyber s?curit? pour les missions essentielles (CME) | Mission Critical Cyber Security (MCCS) R & D pour la d?fense Canada - Valcartier (RDDC Valcartier) | Defence R&D Canada - Valcartier (DRDC Valcartier) 2459 route de la Bravoure Qu?bec QC? G3J 1X5 CANADA Vox?: (418) 844-4000 x4245 Fax?: (418) 844-4538 NAC?: 918V QSDJ Gouvernement du Canada?| Government of Canada