>>>>> "Felix" == Felix Fietkau <[email protected]> writes:
Felix> Please post /proc/meminfo and if you have it, /proc/slabinfo as Felix> well. >> The device represented in the graph (above) is named grank. Look >> in: >> >> http://personaltelco.net/~russell/wgt-memory/ >> >> grank-yyyy-mm-dd-hhmm is /proc/meminfo >> grank-slabinfo-yyyy-mm-dd-hhmm is /proc/slabinfo >> grank-procs-yyyy-mm-dd-hhmm is the output from: Felix> OK, I do see the slab usage increasing significantly, and this Felix> appears to be the issue here, however I need slabinfo dumps Felix> over a longer time period to figure out which slab usage Felix> changes are noise and which ones are relevant I've been collecting /proc/meminfo and /proc/slabinfo for about three weeks now at 10 minute intervals, and they are all available here: http://personaltelco.net/~russell/wgt-memory/ FWIW, I tried to exclude the influence of madwifi by running a recent snapshot (r18567) without a radio installed in a WGT634U (just using ethernet, hostname: clark), and I am still seeing a slow erosion in free+buffers+cache numbers on it. I just started grabbing snapshots of clark's meminfo and slabinfo as well. -- Russell Senior ``I have nine fingers; you have ten.'' [email protected] _______________________________________________ openwrt-devel mailing list [email protected] https://lists.openwrt.org/mailman/listinfo/openwrt-devel
