**** BEGIN LOGGING AT Wed Oct 31 02:59:58 2018 Oct 31 03:42:14 greearb_: needs work on the the ethernet Oct 31 03:42:32 but yes, wifi is excellent Oct 31 04:29:25 Multicast Question: I'm using OpenWRT in plain routing mode (no NAT, no firewall). I have a simple multicast test that joins a group, sends out packets occasionally, and prints whatever it sees (own packets plus packets from other peers) Oct 31 04:30:10 Test nodes on ethernet or wifi see each other (more or less), but no packets pass from eth to wifi or back. (I have the sender set TTL to 5) Oct 31 04:30:48 I'm not seeing any IGMP packets via tcpdump on the router (I expect to see packets to 224.0.0.1 every minute or two) Oct 31 04:31:49 How can I get OpenWRT to forward multicast packets from one interface to another, given that there are hosts on each subnet that are transmitting / joined to the multicast group? Oct 31 07:18:55 build #1069 of cns3xxx/generic is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/cns3xxx%2Fgeneric/builds/1069 Oct 31 09:38:06 build #1045 of lantiq/xrx200 is complete: Success [build successful] Build details are at http://phase1.builds.lede-project.org/builders/lantiq%2Fxrx200/builds/1045 Oct 31 09:41:11 grumble Oct 31 09:44:02 rmilecki: were you able to confirm the fix for ubifs? https://patchwork.ozlabs.org/patch/990017/ Oct 31 10:23:43 hi all Oct 31 10:28:42 o/ Oct 31 10:44:47 welcome back :) Oct 31 12:10:18 CURL 7.62.0 is out https://daniel.haxx.se/blog/2018/10/31/curl-7-62-0-moar-stuff/ Oct 31 12:17:02 xback: thanks Oct 31 12:49:32 I did a custom image of tl-wr841-v8 just with packages luci luci-ssl x2ltpd the rom seems full, cannot connect to HTTPS, is luci not fitting? or is slowly generating https cert ? Oct 31 12:50:42 ssh to it and ffind out? Oct 31 12:52:52 Thans karlp (IRC) . Yes ssh and the l2tp tunnel works, I checked dmesg and logread to see relevant logs, but nothing. What else? Oct 31 12:54:03 is there a uhttpd service listening on the prots you think there is? Oct 31 12:55:09 uhttpd listening in 443 Oct 31 12:55:23 checked with `netstat -tupanl | grep 443` Oct 31 12:55:45 anyone have a suggestion for a cheap-ish AP with QCA 9888 or IPQ4019 chipset, 128MB of storage, 512M RAM preferably Oct 31 12:57:59 ok, I'm going to quit luci-ssl, probably is too much and I don't need it Oct 31 12:58:52 are you firewalled from it? are you connecting from it's wan port or antyhign? Oct 31 13:01:03 so, this was a openwrt 15.0x and it worked well, but I messed up with free space, so I put a openwrt 17.0x with the same etc (yay, doing it remotely and at the moment fine) Oct 31 13:01:24 I don't know, probably there is no space for the https certificate Oct 31 13:07:51 another sysupgrade without luci-ssl, it worked! Oct 31 13:07:54 now luci is available Oct 31 13:46:53 this sysupgrade abort is getting tedious. https://zerobin.net/?90a7ca7c3a68ce2d#+gKhWIY0rFrOUQH7RZe07+6T6YZdC7Hr1GBmMwYhCQk= Oct 31 13:47:09 why does it "always" (too often) fail to term/kill things Oct 31 14:28:00 karlp: hm,, could you figure out why netdata is not killable? Oct 31 14:44:55 it does seem to be netdata, but if I just call /etc/init.d/netdata stop, it alays "just stops" Oct 31 14:45:06 I don't get any indication of there being a problem. Oct 31 14:49:25 purely informational. I'm also using netdata here, and I have not encountered itcould not be stopped on sysupgrade Oct 31 14:49:41 maybe time for https://bugs.openwrt.org/index.php?do=details&task_id=609&string=runlevel :) Oct 31 14:49:55 it always properly terminates in the TERM phase Oct 31 14:50:12 xback: are you using the statsd portion? that's the only change I've got vs stock out of the box config Oct 31 14:50:51 stintel: how does that help? Oct 31 14:51:47 well, it would stop netdata via procd which you just said works fine Oct 31 14:52:03 karlp: I'm using the stock config Oct 31 14:55:32 stintel: how does procd run level translate into that? Oct 31 14:56:07 isn't procd just term/killing anyway? Oct 31 14:56:19 netdata doesn't have any special stop() handling Oct 31 14:56:30 if implemented properly it would do the same as what /etc/init.d/netdata stop would do Oct 31 14:56:45 instead of sending kill signals Oct 31 14:56:50 that's the whole point of implementing this Oct 31 14:57:42 and it should take into account things like term_timeout Oct 31 15:00:31 so, if "that" is implemented, instead of just doing term/kill, it would call the services stop() routine? I mean, ok, that might help domooticz, and some others in general, but it won't make any difference to netdata with no stop() method Oct 31 15:01:13 then how does it "just stop" with /etc/init.d/netdata stop Oct 31 15:01:27 however procd managed services are normalyl stopped Oct 31 15:01:32 ie, there's no special handling Oct 31 15:01:48 I don't know why it normally seems to be netdata that hasn't exited in time for sysupgrade to abort Oct 31 15:48:09 * ldir is stuck on the ground in a plane until 1630 at least, shoulda left 1440 Bored. Very bored. Oct 31 15:51:34 sounds like ryanair :P Oct 31 15:53:47 lol Oct 31 15:54:58 This is tap Portugal but I can’t blame them for the weather Oct 31 16:09:05 ldir: fun Oct 31 16:09:18 i missed my flight this morning Oct 31 16:09:30 700eur later, the last chopper out of saigon Oct 31 16:09:48 was looking like i was staying at one point.. Oct 31 16:15:20 And captain just announced we’re bumped to 1710 Oct 31 16:16:14 yikes Oct 31 16:17:37 So that’s taking off 20min before due to land on a 2.5hr flight hmmmm oh well Oct 31 16:18:52 ldir: yesterday it was like 10 minutes of rain, 10 minutes of sunshine, loop() at the airport, so we were delayed a bit as well Oct 31 16:19:28 i was expecting a takeoff like if i was sitting in a washing machine, but it turned out to be not that bad Oct 31 16:20:57 Ha we seem to be stuck on rain. And when pilot talking of turbulence when we do leave I’m fully expecting the washing machine rollercoaster ride :-) Oct 31 16:21:22 Hi are any of the talks from the openwrt confrence on the web yet? Oct 31 16:22:05 Which conference? Oct 31 16:22:24 * ldir grins and hides Oct 31 16:24:57 * ldir plays a game. Purple Rain Oct 31 16:29:27 And then reads a book how to land an airbus a330 in an emergency and then realises he’s on an a320. How different can it be!? Oct 31 16:30:34 not _that_ much, cockpit layout is similar Oct 31 16:31:44 and it the plane is intact, you'll only need to find an autoland-capable runway Oct 31 16:32:25 Indeed Oct 31 17:04:21 It's not that hard anyway Oct 31 17:08:16 in conditions where passengers would have to consider landing? ok :) Oct 31 17:11:02 Hi, here is an alix and I see this procedure https://wiki.openwrt.org/doc/howto/generic.sysupgrade#openwrt_on_x86 but I don't know if I'm using combined-squash or rootfs-quash (to do the appropriate sysupgrade) -> this is how I know is squashfs -> `VFS: Mounted root (squashfs filesystem) readonly on device 31:0.` Oct 31 17:27:50 ldir: left yet? :) Oct 31 17:28:10 * jwh just got to Munich Oct 31 17:28:22 what a fun day Oct 31 17:28:41 40 mins in the air and ill be home at least Oct 31 17:51:35 what is the difference between 'flash' and NAND flash, ie for the TEW-829DRU Oct 31 17:55:49 hmmm, maybe it doesn't matter..only 256MB RAM on that thing, and I probably need 512M to run comfortably Oct 31 17:56:58 greearb_: It has a small (8MiB) NOR flash for the bootloader and such Oct 31 17:57:15 Small is obviously relative Oct 31 17:57:18 so NAND is normal-ish flash too and can hold root FS etc? Oct 31 18:03:27 greearb_: NAND and NOR are two different kind of persistent storage, and is unrelated to RAM Oct 31 18:03:54 I know that part, but I don't know what is difference between NAND and NOR Oct 31 18:06:00 NOR is quite robust and can be erased very often (>100k times per block) before producing errors, NAND can get flipped bits just by reading a block. So NAND needs special handling to make those bit flips non fatal (by using a different filesystem that can detect that and move data around etc). But NAND is cheaper than NOR. Oct 31 18:07:47 greearb_: NAND needs hand holding, NOR is trivial to use, basically. Oct 31 18:07:55 ok Oct 31 18:08:56 but from a user perspective, there is no difference, exept NAND is faster and usually offers much more space. All the difficult parts are handled by the kernel/filesystem, and you don't need to worry about it (usually) Oct 31 18:09:08 NAND is *much* faster Oct 31 18:09:10 ok, thanks Oct 31 18:09:13 Except ubi is having some.. issues, lately Oct 31 18:09:28 I loved the sysupgrade speed on my WNDR4300 Oct 31 18:09:28 And there is no MLC support, which is beginning to be a problem Oct 31 18:09:38 I crapped UBIFS the other day on my netgear, but don't know why...a reinstall fixed it Oct 31 18:09:49 The speed is quite amazing though, yes, flashing takes all of a few seconds vs hurry up and wait for a minute or three. Oct 31 18:10:06 Monkeh: and there are already TLC NANDs ;D Oct 31 18:10:25 but yeah, MLC NAND is ugly Oct 31 18:10:29 KanjiMonster: Yes, well, enough said about Schrödinger's data.. Oct 31 18:11:25 KanjiMonster: Just wait for QLC, that's sure to be exciting Oct 31 18:11:25 when googling for those hynix part numbers I noticed some of the MLC chips seems to support a command to switch it to SLC mode (which then likely hides the linked pages, effectivly halving the size) Oct 31 18:11:52 There's also some devices with eMMC. Do we have support for eMMC enabled in kernel? Oct 31 18:12:13 also there is a patchset from 2016 for UBI to add SLC emulation by ignoring the paired pages in software **** BEGIN LOGGING AT Wed Oct 31 18:12:35 2018 Oct 31 18:12:53 luaraneda: eMMC doesn't need any special handling as the flash controller abstracts out all the ugly stuff Oct 31 18:12:59 Should just shoot vendors who use MLC or TLC without a suitable controller. Oct 31 18:13:29 * greearb_ trolls alibaba for a 9984 chipset AP with serial port....slow going! Oct 31 18:13:37 MLC in SLC mode is still better than UBI on MLC without proper handling Oct 31 18:13:58 Monkeh: MLC is already a problem with how it manages the pages Oct 31 18:14:02 True, but it's also directly incompatible with anything trying to use it normally Oct 31 18:14:47 KanjiMonster: yeah, but you lose half the capacity. Well, at least it works... Oct 31 18:15:36 luaraneda: I'm aware MLC is a problem already, yes, I sort of brought that up in the first place Oct 31 19:25:25 blogic: back on oct 1 you mentioned some ramips-relevant pinmux changes were in the pipeline, what is that status of that and/or where can i find it? Oct 31 20:52:13 has arrived back in good old, ooops I mean F**KING INCOMPETENT England, where my cab company booking has decided not to keep track of the flight and not turn up. Oct 31 20:52:52 so I have another 30+ min wait for the cab to arrive. Oct 31 20:53:14 ldir: welcome back :-) Oct 31 20:53:42 lol Oct 31 20:53:54 * ldir is in kill mode! :-) Oct 31 21:18:05 lol Oct 31 21:52:28 ldir: no uber? Oct 31 21:55:28 cant believe i made it back before you, i missed my flight *and* had to take 3 to get back (for nothing as it turns out as movers arent here) Nov 01 01:11:33 Is there a make target like menuconfig where you can look at / change the linux kernel configuration settings? Nov 01 01:13:46 make kernel_menuconfig. Too easy. **** ENDING LOGGING AT Thu Nov 01 03:00:00 2018