According to issues discussed over the past year or so on github, seems so - that at least some models and/or OSes and/or chipsets have a problem with libusb interrupt mode; more often failing during device discovery than after some successful uptime. I don't think there was reliable statistics to incriminate certain situations enough to hardcode some preferences however. Maybe it could help generally to at least log the suggestion for pollonly (or interruptonly) when the preferred method fails. Or at best try to change the preference on the fly if not bolted by explicit config. Question to community: would such change (especially if also of default behavior) be beneficial? Jim On Sat, Jun 4, 2022, 19:16 Selva Nair via Nut-upsuser < nut-upsuser at alioth-lists.debian.net> wrote:> Ubuntu 22.04 > NUT 2.7.4 > CYberPower CP1500AVRLCDa using usbhid-ups > > Hello, > > The above combination works well for a few minutes -- including shutdown on > pulling power or on upsmon -c fsd, notifications etc. Until the driver > starts reporting error in libusb_get_string, kernel messages on > USDEVFS_CONTROL failures and upsmon logs 'Data stale' > > It seems the UPS does not support interrupt transfer (or is buggy). On > adding pollonly=enabled to ups.conf, everything appears to be working > fine again. > > Is this a known issue? I have seen a number of success reports on this > model, none mentioning the need for pollonly, so wondering whether I > should troubleshoot this further. > > Thanks, > > Selva > > _______________________________________________ > Nut-upsuser mailing list > Nut-upsuser at alioth-lists.debian.net > https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/nut-upsuser >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://alioth-lists.debian.net/pipermail/nut-upsuser/attachments/20220606/ef027e96/attachment.htm>
Dave Williams
2022-Jun-06 16:15 UTC
[Nut-upsuser] Cyber power CP1500AVR required pollonly?
IMHO adding to the logging to suggest pollonly would be a good start. I wouldnt advocate changing on the fly - thats likely to lead to instability or confusion. On my CP900EPFCLCD I need to run pollonly albeit when connected to old USB 1.5 ports. I thought it was the port not the UPS that was the problem - maybe I was jumping to conclusions. Dave On 17:30, Mon 06 Jun 22, Jim Klimov via Nut-upsuser wrote:> According to issues discussed over the past year or so on github, seems so > - that at least some models and/or OSes and/or chipsets have a problem with > libusb interrupt mode; more often failing during device discovery than > after some successful uptime. > > I don't think there was reliable statistics to incriminate certain > situations enough to hardcode some preferences however. > > Maybe it could help generally to at least log the suggestion for pollonly > (or interruptonly) when the preferred method fails. Or at best try to > change the preference on the fly if not bolted by explicit config. > Question to community: would such change (especially if also of default > behavior) be beneficial? > > Jim > > On Sat, Jun 4, 2022, 19:16 Selva Nair via Nut-upsuser < > nut-upsuser at alioth-lists.debian.net> wrote: > > > Ubuntu 22.04 > > NUT 2.7.4 > > CYberPower CP1500AVRLCDa using usbhid-ups > > > > Hello, > > > > The above combination works well for a few minutes -- including shutdown on > > pulling power or on upsmon -c fsd, notifications etc. Until the driver > > starts reporting error in libusb_get_string, kernel messages on > > USDEVFS_CONTROL failures and upsmon logs 'Data stale' > > > > It seems the UPS does not support interrupt transfer (or is buggy). On > > adding pollonly=enabled to ups.conf, everything appears to be working > > fine again. > > > > Is this a known issue? I have seen a number of success reports on this > > model, none mentioning the need for pollonly, so wondering whether I > > should troubleshoot this further. > > > > Thanks, > > > > Selva > > > > _______________________________________________ > > Nut-upsuser mailing list > > Nut-upsuser at alioth-lists.debian.net > > https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/nut-upsuser > >> _______________________________________________ > Nut-upsuser mailing list > Nut-upsuser at alioth-lists.debian.net > https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/nut-upsuser
Hi, Thanks for the followup. A hint about pollonly option in the logs would be very helpful. Automatically changing the mode could be hard to get right and may not be the proper thing to do in some situations. It would also make troubleshooting harder if the driver uses options contrary to what is in ups.conf. Thanks, Selva On Mon, Jun 6, 2022 at 11:30 AM Jim Klimov <jimklimov+nut at gmail.com> wrote:> > According to issues discussed over the past year or so on github, seems so - that at least some models and/or OSes and/or chipsets have a problem with libusb interrupt mode; more often failing during device discovery than after some successful uptime. > > I don't think there was reliable statistics to incriminate certain situations enough to hardcode some preferences however. > > Maybe it could help generally to at least log the suggestion for pollonly (or interruptonly) when the preferred method fails. Or at best try to change the preference on the fly if not bolted by explicit config. > Question to community: would such change (especially if also of default behavior) be beneficial? > > Jim > > On Sat, Jun 4, 2022, 19:16 Selva Nair via Nut-upsuser <nut-upsuser at alioth-lists.debian.net> wrote: >> >> Ubuntu 22.04 >> NUT 2.7.4 >> CYberPower CP1500AVRLCDa using usbhid-ups >> >> Hello, >> >> The above combination works well for a few minutes -- including shutdown on >> pulling power or on upsmon -c fsd, notifications etc. Until the driver >> starts reporting error in libusb_get_string, kernel messages on >> USDEVFS_CONTROL failures and upsmon logs 'Data stale' >> >> It seems the UPS does not support interrupt transfer (or is buggy). On >> adding pollonly=enabled to ups.conf, everything appears to be working >> fine again. >> >> Is this a known issue? I have seen a number of success reports on this >> model, none mentioning the need for pollonly, so wondering whether I >> should troubleshoot this further. >> >> Thanks, >> >> Selva >> >> _______________________________________________ >> Nut-upsuser mailing list >> Nut-upsuser at alioth-lists.debian.net >> https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/nut-upsuser