so try on the internal flash drive. you will also get an error.
for example, there is no usb port on 4011. And there is this problem.
I don’t run containers on internal memory that’s not that easy to replace (if even possible at all), sorry.
Yeah it’s definitely still a problem for me hey.. happening after each update, I’m not just making things up here.
Here’s exactly what I use (except passwords changed obviously):
/container mounts add dst=/data name=nginxpm-app-data src=/container/nginxpm-app/data
/container mounts add dst=/etc/letsencrypt name=nginxpm-app-letsencrypt src=/container/nginxpm-app/letsencrypt
/container mounts add dst=/var/lib/mysql name=nginxpm-db-mysql src=/container/nginxpm-db/mysql
/container envs add key=DB_MYSQL_HOST name=nginxpm-app value=172.18.0.3
/container envs add key=DB_MYSQL_PORT name=nginxpm-app value=3306
/container envs add key=DB_MYSQL_USER name=nginxpm-app value=myuser
/container envs add key=DB_MYSQL_PASSWORD name=nginxpm-app value=somepass
/container envs add key=DB_MYSQL_NAME name=nginxpm-app value=nginxpm
/container envs add key=MYSQL_ROOT_PASSWORD name=nginxpm-db value=somepass
/container envs add key=MYSQL_DATABASE name=nginxpm-db value=nginxpm
/container envs add key=MYSQL_USER name=nginxpm-db value=myuser
/container envs add key=MYSQL_PASSWORD name=nginxpm-db value=“somepass”
/container add envlist=nginxpm-app interface=veth1-nginxpm-app logging=yes mounts=nginxpm-app-data,nginxpm-app-letsencrypt remote-image=jc21/nginx-proxy-manager:latest
/container add envlist=nginxpm-db interface=veth2-nginxpm-db logging=yes mounts=nginxpm-db-mysql remote-image=jc21/mariadb-aria
“/container/” is not on on an external storage - it’s a directory which gets created by routeros when creating the containers on the built-in storage.
since the last 2 updates (from 7.6beta4->7.6beta6, and beta7.6beta6->7.6beta7), I made sure to completely rebuild the containers when I noticed the permissions problem. That means I deleted all the container instances, envlists and mountpoints config, and the whole /containers directory, and started over. But that didn’t help.
So I’m not too sure why I seem to be the only person still experiencing this issue. Maybe something’s wrong with my CHR’s filesystem ? Why does it only break on update though?
I have a ticket open, SUP-92866. Hoping for the best.. & Sorry for flooding the thread a bit with my complaint xD
Any reason why you’re not setting a root-dir for your containers? where do they get created?
Post an output of /file/print
Yeah I’m wondering if that might not be the cause. RouterOS generates a random ID & root-dir automatically, which I assumed was by design. This is retained after the upgrade, but perhaps there’s an issue with permissions if root-dir isn’t manually specified like you say?
[admin@chr] > /file/print
Columns: NAME, TYPE, SIZE, CREATION-TIMENAME TYPE SIZE CREATION-TIME
0 supout.rif .rif file 614.5KiB sep/20/2022 03:54:34
1 us2.rsc script 43.8KiB aug/30/2022 08:08:55
2 8771505e-94ad-41a8-b2e6-9ba67a2dcd05 container store sep/19/2022 08:16:06
3 8a21452c-17d5-4b41-b688-ae88797ceb13 container store sep/19/2022 08:14:46
4 container directory sep/19/2022 08:16:15
5 container/nginxpm-app directory sep/19/2022 08:16:06
6 container/nginxpm-app/data container store sep/19/2022 08:16:06
7 container/nginxpm-app/letsencrypt container store sep/20/2022 04:16:28
8 container/nginxpm-db directory sep/19/2022 08:16:15
9 container/nginxpm-db/mysql container store sep/19/2022 08:16:18
[admin@chr] >
The bold lines represent the automatically generated root dirs
Nah, doubt it, I thought they might interfere with your mounts but it doesn’t seem to be the case. I’ll try testing AdGuardHome in a CHR and see what I come up with.
But since it’s a CHR can’t you add to it another disk for containers?
It’s an actual cloud-hosted CHR which comes with 50GB disk included the package (plenty enough for some small containers without needing to add another disk which I’d have to pay more for)…
But support have just gotten back to me with good news! – “Thank you for the report! We have managed to reproduce the issue locally in our labs and look forward to fixing it on upcoming RouterOS versions, unfortunately, I cannot provide a release date now. Best regards,” – So hopefully we can put this one to rest now 
Thanks for all the suggestions & feedback
You are not the only one having this problem. I already write about every release about it.
AdGuardHome has the same problem.
Nice, I’ve managed to reproduce it too in the meantime.
I’ve installed CHR 7.6beta6, resized the main disk, installed AdGuardHome, upgraded to 7.6beta7 → problem.
I’ve installed CHR 7.6beta6, added an extra disk, installed AdGuardHome on the 2nd disk, upgraded to 7.6beta7 → ok.
So yes there still is a bug when you run containers on internal storage, probably same thing @tpedko is experiencing on his RB4011.
I don’t have any MikroTik router with enough internal storage to sacrifice for a test.
Yes it’s a problem with CHR instances, but with actual routers you SHOULD NOT run containers on their internal memory, as if the warning on the documentation page isn’t enough.
On 5009 the same problem.
I think they can’t fix the problem and that’s why the documentation has a warning.
No, the documentation has a warning so that you don’t wear out the internal memory with container garbage and then cry that your router died too soon.
This bug regarding mount permissions was recently fixed in 7.6beta4 but it only applies to containers sitting outside of the internal memory, it seems. It’ll get fixed, like the reply that @fragtion posted above says.
The warning that you shouldn’t put garbage on your internal precious memory will always be there.
I don’t have this issue, using CCR2216 with 58 bgp peers…sessions are stable…
DOM/DDM on my RB760iGS still not work.
What’s new in 7.6beta8 (2022-Sep-21 09:20):
Important note!!!
Version not recommended on TILE and RB5009 devices if MACsec is used;
Changes in this release:
*) bgp - added support for BGP advertisement displaying (CLI only);
*) certificate - fixed SHA1 certificate name lookup;
*) dhcpv4-server - fixed matcher functionality;
*) ethernet - added “5Gbps” option for speed setting;
*) firewall - fixed usage of “netmap” action for IPv6 source NAT (CLI only);
*) lte - added periodic IPv6 RS to trigger IPv6 adress acquisition for non-MBIM modems;
*) macsec - fixed interface after Ethernet link down;
*) macsec - fixed interface statistics and missing properties;
*) macsec - fixed interface status;
*) macsec - fixed multiple interface creation on different Ethernet ports
*) macsec - removed interface from SMIPS devices;
*) ospf - added SHA hashing for authentication;
*) queue - improved stability for CAKE type queues;
*) snmp - improved retrieval of routing related OID’s;
*) sstp - added IPv6 support (CLI only);
*) switch - improved traffic forwarding at 5Gbps rate for 98DX8525, 98DX4310 switches;
*) tile - improved system stability when processing packets;
*) webfig - fixed hex input for “Host Uniq” field;
*) winbox - added “Active” prefix for current remote and local session ID fields for L2TP-Ether interfaces;
*) winbox - added “address-list” parameter under “IP/DNS/Static” menu;
*) winbox - added MACsec support;
*) winbox - added “type” and “status-report-request” parameters under “Tools/SMS” menu;
*) winbox - changed “uptime” parameter format when using the wifiwave2 package;
*) wireless - disallowed using “default” as scan list or channel names;
New /routing/stats/adverts menu, feedback, and suggestions are welcome.
Q:
*) wireless - disallowed using “default” as scan list or channel names
What is this specifically about ?
How does this relate to Connect List ?
OK, you can drop this question.
Completely unrelated.
/stats/adverts menu, feedback, and suggestions are welcome.
Need adverts prefix send peer bgp
DOM/DDM on my RB760iGS still not work.
Is there any fix mentioned in the changelog and it doesn’t work?