Using the bufferbloat test I am now getting about 80ms more latency on download. Uploads are still low and the same as before.
No config changes. Just upgrading from 22.03.2 to 22.03.3.
Oh, I didn't know of the existence of that site (just the BroadbandReports one, which I haven't gotten to work at all in recent years), so have no before and after, but it looks good to me, with Download only moving from 29 to 36 (overall grade B), which is more in line with what I recall from my manual ping tests before.
But I re-ran the manual ping test and still do see it soaring into the 70s and 80s, so it wasn't a fluke.
(a) thanks for pushing the fix
(b) Since the release info page is a wiki page, I presume it's possible to add and remove items or update them as the situation changes. But perhaps the project doesn't think the known issues section is the place for this sort of known regression. So then where should the project document problems like this that users would want to know about?
Thank you for answering my questions and also confirming the success on your device with snapshots and kernel 5.15.
My situation is from question 2, so mostly wireless but I do have 1 wired device therefore probably not affected.
It seems as though 22.03.x series would be OK for some users and not OK for others depending on their network setup.
However, staying on builds with known security issues is a bigger problem. So moving to snapshots or going back to 21.02.x (supported until April) is essential for the time being.
3 Likes
I use a WRT3200ACM. If I now want to take all security updates without going to a snapshot build I could also build my own image for the WRT3200ACM via the image builder (openwrt-imagebuilder-22.03.3-mvebu-cortexa9.Linux-x86_64.tar.xz) or is there something against it? Would then update from 22.03.2 to 22.03.3.
So this seems a bit paradoxical, no images but the image builder is extent.
Edit: point being, if the IB allows you to assemble an image for one of the disabled devices, it is the same kernel, so should exhibit the same behaviour that forced the devices to be marked disabled.
For me I don't see any problems with switch. But I'm using only networks separated by vlans on the switch.
Is this thru wifi, or eth? Does enabling SQM still bring it down?
Edit: Sorry, missed that this is with SQM in your first post.
Hi.
I updated the MR8300 page last week, to inform about the necessary change for latest snapshot and future 23.xx. There have been also some behavior changes in flashing, that have been backported to 22.03.3. For what I remember from the thread we used (https://github.com/openwrt/openwrt/pull/11405) it is meant to use a factory image with sysupgrade, or even OEM firmware. Nevertheless I haven't tested this feature. Best to do is to ask directly to the author : @guidosarducci
Testing through ethernet. I wanted to see if this was just on my arch (mips64_octeonplus) or is this across the board.
Thanks. I installed the snapshot for my mvebu Linksys 1900ACS router. Before that, I tried a test I read along the way (ping -i 0.1 <router name/addr> and noticed that was directed to both of the machines connected to the router. But now, after installing the snapshot (r21729-89eb8b50d1), only the router responds to the ping; which is the way it should be, I believe. Thanks for the link to the snapshot.
2 Likes
Enter your target into the snapshot dropdown and grab a build from master, these include kernel 5.15 which fixes the switch on mvebu:
https://firmware-selector.openwrt.org/?version=SNAPSHOT
Here are the packages I install via SSH on a clean build which you might want use (lot of storage stuff because I use Samba, it's very fast 120MB/s on our targets):
Summary
opkg update && opkg install luci luci-ssl irqbalance luci-app-advanced-reboot luci-app-sqm luci-app-simple-adblock luci-app-upnp luci-app-samba4 kmod-usb3 kmod-ata-marvell-sata kmod-usb-storage kmod-usb-storage-uas block-mount usbutils mount-utils luci-app-hd-idle kmod-fs-exfat nano nmon htop iperf3
1 Like
yes i have the same problem for x86-64
Got a question to you all. Why there's no package bind-server available for 22.03.3? Anyone answers for this? Thank you for answering.
Asked and answered, read back; I assume the search button works.
2 Likes
OK....This mvebu switch thing has got me spooked
I usually leave the "world of snapshots" stuff to you qualified guys but it sounds like we mvebu owners need to take some action this time.
I use the imagebuilder to compile my flash images but I have not ever used one for a snapshot.
I DL'd "openwrt-imagebuilder-mvebu-cortexa9.Linux-x86_64.tar.xz" from the Index of (root) / snapshots / targets / mvebu / cortexa9 / and added my packages and "files" to it and ran it as usual.
It compiled normally except for one package.
Flashed a WRT3200ACM with the image and it is up and running with luci.
Status overview shows:
Firmware Version OpenWrt SNAPSHOT r21729-89eb8b50d1 / LuCI Master git-22.361.69865-deed682
Kernel Version 5.15.86
Is there anything else that needs to be done to complete a snapshot install?
Will system software update properly?k Or not at all?
1 Like
I'm not sure what packages you installed with master snapshot to know if there is anything else needed (e.g. advanced-reboot, sqm, adblock, samba, etc.), but if you're up and running it sounds like you're good.
For updates you can just flash a new sysupgrade snapshot anytime, they are built almost daily. My current build has been running 3 weeks uptime, keeping it going until the new GCC 12 commit gets a week or so I'll flash a new one if it's using that. Usually I just flash a new one monthly or so.
I just installed the SNAPSHOT (r21729-89eb8b50d1) and it seems to run fine on my WRT3200ACM. I had to install some luci and other packages. Am I on the safe side for now? Should I reinstall as new snapshots come out?
You should be good for a bit, can keep an eye on the github for needed changes, or just install a sysupgrade snapshot monthly or so until the next release which should include 5.15 by default.
1 Like