Re: Poweroff on Fedora 37
On 5/12/23 18:48, Jeffrey Walton wrote: On Fri, May 12, 2023 at 8:04 PM Samuel Sieb wrote: On 5/12/23 16:20, Jeffrey Walton wrote: On Fri, May 12, 2023 at 5:10 PM Joe Zeff wrote: On 05/12/2023 02:58 PM, Jeffrey Walton wrote: So does the hostname command, but it no longer works, either. I just tried the basic command, and it worked just fine. What doesn't it do for you that it's supposed to? Set the hostname using the command. It works for me, but why would you want to do that? It's not a permanent change anyway. I'd be interested to know how it works for you. Setting a host's name using hostname is _the way_ to set a host name. Or it was until Poettering came along. Maybe you should read the man page for hostname. If you want a permanent change, you set it in /etc/hostname (or now you can use "hostnamectl"). But the hostname command still works for temporary changes. You didn't explain what you're trying to do and what's not working, although you should start a new thread because this one has gone off track. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
On Fri, May 12, 2023 at 8:04 PM Samuel Sieb wrote: > > On 5/12/23 16:20, Jeffrey Walton wrote: > > On Fri, May 12, 2023 at 5:10 PM Joe Zeff wrote: > >> > >> On 05/12/2023 02:58 PM, Jeffrey Walton wrote: > >>> So does the hostname command, but it no longer works, either. > >> > >> I just tried the basic command, and it worked just fine. What doesn't > >> it do for you that it's supposed to? > > > > Set the hostname using the command. > > It works for me, but why would you want to do that? It's not a > permanent change anyway. I'd be interested to know how it works for you. Setting a host's name using hostname is _the way_ to set a host name. Or it was until Poettering came along. You don't have to take my word on it. W. Richard Stevens writes prodigiously about these topics in his books. The hostname program is covered in APUE p. 155. Jeff ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Any x86_64 Fedora Tablets out there?
On 5/12/23 17:44, Allan via users wrote: On Thu, 11 May 2023 17:01:35 -0700 ToddAndMargo via users wrote: On 5/11/23 15:35, Allan via users wrote: On Mon, 8 May 2023 18:51:54 +0200 Mario Marietto wrote: my newborn interest is already over. Well, we do try to support a few of them in the Mobility SIG I'm also interested to find a tablet where I can install Fedora,but that tablet should be based on ARM. https://pine64.com/product-category/tablets/ We will try to support these, when they start sending them out. Allan. Hi Allen, Awesome! The pine64 is an ARM tablet. Correct. That said, it is completely new on the market - and not yet fully supported by mainline kernel and thereby Fedora. It is expected for users (in the beginning) to have some super-user skills. The Pinetab2 is also a low-power SoC - very far from a powerful SoC as you generally see them on x86 laptops. I have one Windows app that required an intel processor that I need to run in Wine (it does on my Fedora37). I presume that is out with ARM. Well, I can't really answer that, as I have never used wine. To my big surprise though, wine is available in Fedoras repos for aarch64. I suspect though, that it will require Windows aarch64 binaries to be able to run them. An alternative could be an emulator like Box64, that can run x86_64 files on aarch64 - but that will be _extremely_ slow on this device. But, if I can run Android apps, that same app also has an ARM Android version, which works better than the windows version! Fedora has Waydroid in repos, that can run Android apps in a container. Again generally reported so be slow on such low-power devices. Your take? Will I be able to run ARM android apps? Possibly. Also, maybe a dumb question, but does ARM Fedora also come with SE Linux? Yes. And, can I run MATE and/or Xfce on the ARM version? You can run whatever you like. Allan. Thank you! ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Any x86_64 Fedora Tablets out there?
On Thu, 11 May 2023 17:01:35 -0700 ToddAndMargo via users wrote: > On 5/11/23 15:35, Allan via users wrote: > > On Mon, 8 May 2023 18:51:54 +0200 > > Mario Marietto wrote: > > > >> > >> my newborn interest is already over. > > > > Well, we do try to support a few of them in the Mobility SIG > > > I'm also interested to find a tablet where I can install > Fedora,but that tablet should be based on ARM. > > > > https://pine64.com/product-category/tablets/ > > > > We will try to support these, when they start sending them out. > > > >Allan. > > Hi Allen, > > Awesome! > > The pine64 is an ARM tablet. Correct. That said, it is completely new on the market - and not yet fully supported by mainline kernel and thereby Fedora. It is expected for users (in the beginning) to have some super-user skills. The Pinetab2 is also a low-power SoC - very far from a powerful SoC as you generally see them on x86 laptops. I have one Windows > app that required an intel processor that I > need to run in Wine (it does on my Fedora37). > I presume that is out with ARM. Well, I can't really answer that, as I have never used wine. To my big surprise though, wine is available in Fedoras repos for aarch64. I suspect though, that it will require Windows aarch64 binaries to be able to run them. An alternative could be an emulator like Box64, that can run x86_64 files on aarch64 - but that will be _extremely_ slow on this device. > But, if I can run Android apps, that same app also > has an ARM Android version, which works better > than the windows version! Fedora has Waydroid in repos, that can run Android apps in a container. Again generally reported so be slow on such low-power devices. > Your take? Will I be able to run ARM android apps? Possibly. > Also, maybe a dumb question, but does ARM Fedora > also come with SE Linux? Yes. > And, can I run MATE and/or Xfce on the ARM version? You can run whatever you like. Allan. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: F38/Xfce - background DNF checks still going
On Fri, 2023-05-12 at 13:08 -0400, Robert Moskowitz wrote: > Looks like I am going to have to just uninstall dhf-dragora. > > Even after > > systemctl disable dnf-makecache.timer > > Still got > > dnf update > Last metadata expiration check: 0:18:17 ago on Fri 12 May 2023 12:45:30 > PM EDT. > > ARGH!!! wasting cycles and bandwidth for what? So your command line tool responds quicker? Mostly it'll parse the prefetched data, rather than have to get all the metadata when you run it. Unless, I'm guessing, it decides that too much time has passed since it downloaded metadata and when you ran the dnf tool. I removed dnf-dragora, and dnf-makecache.timer still exists. They're different things. dnf-dragora is that task bar thing that pops up telling you there's packages available to update, and you can fire off the updates from it, and a GUI tool for installing packages. I removed it, I don't use them at all. dnf-makecache.timer simply keeps the dnf metadata refreshed (which probably isn't essential). It's a part of the main dnf RPM, so I don't think you can get rid of it. Try masking it if you want. -- NB: All unexpected mail to my mailbox is automatically deleted. I will only get to see the messages that are posted to the list. The following system info data is generated fresh for each post: uname -rsvp Linux 6.2.14-100.fc36.x86_64 #1 SMP PREEMPT_DYNAMIC Mon May 1 00:54:35 UTC 2023 x86_64 ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
On 5/12/23 16:20, Jeffrey Walton wrote: On Fri, May 12, 2023 at 5:10 PM Joe Zeff wrote: On 05/12/2023 02:58 PM, Jeffrey Walton wrote: So does the hostname command, but it no longer works, either. I just tried the basic command, and it worked just fine. What doesn't it do for you that it's supposed to? Set the hostname using the command. It works for me, but why would you want to do that? It's not a permanent change anyway. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
On 05/12/2023 05:20 PM, Jeffrey Walton wrote: On Fri, May 12, 2023 at 5:10 PM Joe Zeff wrote: On 05/12/2023 02:58 PM, Jeffrey Walton wrote: So does the hostname command, but it no longer works, either. I just tried the basic command, and it worked just fine. What doesn't it do for you that it's supposed to? What is the exact command you used, and did you do it as root? ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
On Fri, May 12, 2023 at 5:10 PM Joe Zeff wrote: > > On 05/12/2023 02:58 PM, Jeffrey Walton wrote: > > So does the hostname command, but it no longer works, either. > > I just tried the basic command, and it worked just fine. What doesn't > it do for you that it's supposed to? Set the hostname using the command. Jeff ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
If the switch is dumb enough and won't disable a port if it sees the same mac address on 2 ports then linux bonding active-active round-robin has worked for me before. If the switch is smart enough it will disable one of the 2 ports. This is using dumb bonding no LAG, and the switch will bounce the mac between the ports quickly. On Fri, May 12, 2023 at 4:39 PM Thomas Cameron wrote: > > On 5/12/23 16:32, Samuel Sieb wrote: > > If you have two gigabit interfaces and a managed switch, you can also > > team the interfaces for double the bandwidth. Still much cheaper than > > going to 10Gb. > > Sadly, right now I just have an unmanaged switch. I would either have to > upgrade to a new switch or buy a new 10Gb switch. > > If I buy a new switch, I may as well spend a little more for MUCH higher > throughput. > > Thomas > ___ > users mailing list -- users@lists.fedoraproject.org > To unsubscribe send an email to users-le...@lists.fedoraproject.org > Fedora Code of Conduct: > https://docs.fedoraproject.org/en-US/project/code-of-conduct/ > List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines > List Archives: > https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org > Do not reply to spam, report it: > https://pagure.io/fedora-infrastructure/new_issue ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
On 5/12/23 14:38, Chris Adams wrote: Once upon a time, Samuel Sieb said: If you have two gigabit interfaces and a managed switch, you can also team the interfaces for double the bandwidth. Still much cheaper than going to 10Gb. You do not get double the bandwidth from a LAG, except under the most ideal circumstances; you probably get an increase in overall traffic, but usually not at all for something like NFS (which uses a single TCP socket for communication). LAGs don't balance or round-robin traffic; they hash some selection of packet info (sometimes just source/destination MAC, sometimes adding IP, sometimes also TCP/UDP src/dest port) and select a LAG member to use based on the hash. All packets of a single stream go down the same LAG member, because otherwise you introduce jitter and out-of-order packet arrival. Yes, I just looked it up now. That's unfortunate. But it will still likely work for my purposes anyway (multiple devices accessing a higher than 1Gbs internet connection). ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
On 5/12/23 16:32, Samuel Sieb wrote: If you have two gigabit interfaces and a managed switch, you can also team the interfaces for double the bandwidth. Still much cheaper than going to 10Gb. Sadly, right now I just have an unmanaged switch. I would either have to upgrade to a new switch or buy a new 10Gb switch. If I buy a new switch, I may as well spend a little more for MUCH higher throughput. Thomas ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
Once upon a time, Samuel Sieb said: > If you have two gigabit interfaces and a managed switch, you can > also team the interfaces for double the bandwidth. Still much > cheaper than going to 10Gb. You do not get double the bandwidth from a LAG, except under the most ideal circumstances; you probably get an increase in overall traffic, but usually not at all for something like NFS (which uses a single TCP socket for communication). LAGs don't balance or round-robin traffic; they hash some selection of packet info (sometimes just source/destination MAC, sometimes adding IP, sometimes also TCP/UDP src/dest port) and select a LAG member to use based on the hash. All packets of a single stream go down the same LAG member, because otherwise you introduce jitter and out-of-order packet arrival. Also, 10G has lower latency than 1G, which helps NFS performance as well. -- Chris Adams ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
On 5/12/23 07:41, Roger Heflin wrote: At home I question the value of it. You might even simply test your nfs home setup and see if you can even get close to Gbit speeds, unless you have a newer machine and a good underlying disk setup you probably aren't going to get there. And you also need a decent Gbit card, but if you do not have a decent gbit card, you can buy a used enterprise grade dual gbit card for around $25 (branded Dell or branded HPE). If you have two gigabit interfaces and a managed switch, you can also team the interfaces for double the bandwidth. Still much cheaper than going to 10Gb. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
Once upon a time, Ian Pilcher said: > For short runs, you can use DAC cables; no need for separate SFP+ > modules. FS.com has 2m 10G DAC cables for $14. It's a trade-off... cards that take SFP+s tend to cost more than cards with just an RJ-45 jack. I do have a DAC cable between my router and switch (because both are SFP+), but then I have RJ-45 SFP+s in the switch for the computer ports (because the computers are just RJ-45). > (Just make sure that their coded properly for the hardware that you're > going to use them with.) This is sooo stupid. Even as network vendors have mostly backed off this crap, Intel's drivers in the Linux kernel enforce vendor coding. The driver for older chips has a module option to disable it, but the more recent driver for newer chips doesn't even have that. I was very surprised when I got bit by this at a previous job - we tended to get FiberStore "generic" coded modules, which work just fine in equipment from multiple vendors, but then newer Dell servers with newer Intel 10G chips rejected them. We shipped out servers to a customer, who only after shipment said they wanted 10G rather than 1G connections, and asked if they could install an SFP+ they had on hand (it was a telephone company ISP, they had lots), and we said "sure" and proceeded to go through lots of troubleshooting, with them swapping modules, before it got to me and I checked and saw the Linux kernel rejecting the modules. Embarrassing for us. -- Chris Adams ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Firefox - detection of exiting tabs with URLs
On Fri, 2023-05-12 at 23:27 +0930, Tim via users wrote: > On Fri, 2023-05-12 at 11:38 +0100, Patrick O'Callaghan wrote: > > Looks like you should report this upstream. I doubt it's a Fedora > > bug. > > > > poc > > POC, you didn't trim your quotes! ;-) > > I had to wait a long time to get that joke in. Mea culpa :-) poc ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: F38/Xfce - background DNF checks still going
On 5/12/23 14:08, Robert Moskowitz wrote: On 5/12/23 13:20, Samuel Sieb wrote: On 5/12/23 10:08, Robert Moskowitz wrote: On 5/11/23 05:56, Patrick O'Callaghan wrote: On Thu, 2023-05-11 at 19:13 +0930, Tim via users wrote: On Wed, 2023-05-10 at 16:41 -0400, Robert Moskowitz wrote: I thought I disabled dnfdragona from doing its 3? hour check for anything new. But whenever I do a "dnf update" I see it starting with a message like: If you don't use dnf-dragora you don't need it installed. I always disabled it, then finally decided to remove it the other day. I didn't want its interrupting notices, and never used to it to install anything. I always did a "dnf update" on the command line at a time of my own choosing. Same here. Looks like I am going to have to just uninstall dhf-dragora. Even after systemctl disable dnf-makecache.timer Still got dnf update Last metadata expiration check: 0:18:17 ago on Fri 12 May 2023 12:45:30 PM EDT. Look in the logs to see what's triggering that. unless I am looking in the wrong log, I am not seeing what is the trigger: # tail dnf.log -n20 That's the wrong log. The dnf log won't tell you what started dnf. You need to look in the journal. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
On 5/12/23 13:58, Jeffrey Walton wrote: On Fri, May 12, 2023 at 4:52 PM Joe Zeff wrote: On 05/12/2023 02:36 PM, Jeffrey Walton wrote: Unfortunately, shutdown and reboot are not Posix commands, so they don't have specified behavior: https://pubs.opengroup.org/onlinepubs/9699919799/idx/utilities.html That may be, but shutdown, at least has a man page. So does the hostname command, but it no longer works, either. It would not bother me if systemd reimplemented the commands, and then replaced a command with a systemd equivalent, like hostname calling hostnamectl. But the systemd folks just broke it without a care. What isn't working? I've never had any problems with any of those commands. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
On 05/12/2023 02:58 PM, Jeffrey Walton wrote: So does the hostname command, but it no longer works, either. I just tried the basic command, and it worked just fine. What doesn't it do for you that it's supposed to? ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: F38/Xfce - background DNF checks still going
On 5/12/23 13:20, Samuel Sieb wrote: On 5/12/23 10:08, Robert Moskowitz wrote: On 5/11/23 05:56, Patrick O'Callaghan wrote: On Thu, 2023-05-11 at 19:13 +0930, Tim via users wrote: On Wed, 2023-05-10 at 16:41 -0400, Robert Moskowitz wrote: I thought I disabled dnfdragona from doing its 3? hour check for anything new. But whenever I do a "dnf update" I see it starting with a message like: If you don't use dnf-dragora you don't need it installed. I always disabled it, then finally decided to remove it the other day. I didn't want its interrupting notices, and never used to it to install anything. I always did a "dnf update" on the command line at a time of my own choosing. Same here. Looks like I am going to have to just uninstall dhf-dragora. Even after systemctl disable dnf-makecache.timer Still got dnf update Last metadata expiration check: 0:18:17 ago on Fri 12 May 2023 12:45:30 PM EDT. Look in the logs to see what's triggering that. unless I am looking in the wrong log, I am not seeing what is the trigger: # tail dnf.log -n20 2023-05-12T16:25:47-0400 DEBUG reviving: 'updates' can be revived - metalink checksums match. 2023-05-12T16:25:49-0400 DEBUG updates: using metadata from Fri 12 May 2023 12:04:00 AM EDT. 2023-05-12T16:25:49-0400 DEBUG countme: no event for updates-modular: window already counted 2023-05-12T16:25:49-0400 DEBUG reviving: 'updates-modular' can be revived - metalink checksums match. 2023-05-12T16:25:49-0400 DEBUG updates-modular: using metadata from Tue 20 Feb 2018 02:18:14 PM EST. 2023-05-12T16:25:49-0400 DEBUG reviving: 'rpmfusion-free' can be revived - metalink checksums match. 2023-05-12T16:25:49-0400 DEBUG rpmfusion-free: using metadata from Fri 14 Apr 2023 07:37:11 AM EDT. 2023-05-12T16:25:50-0400 DEBUG reviving: failed for 'rpmfusion-free-updates', mismatched sha256 sum. 2023-05-12T16:25:50-0400 DEBUG repo: downloading from remote: rpmfusion-free-updates 2023-05-12T16:25:52-0400 DEBUG rpmfusion-free-updates: using metadata from Fri 12 May 2023 02:06:31 PM EDT. 2023-05-12T16:25:52-0400 DEBUG reviving: 'rpmfusion-nonfree' can be revived - metalink checksums match. 2023-05-12T16:25:52-0400 DEBUG rpmfusion-nonfree: using metadata from Fri 14 Apr 2023 08:02:51 AM EDT. 2023-05-12T16:25:52-0400 DEBUG reviving: 'rpmfusion-nonfree-updates' can be revived - metalink checksums match. 2023-05-12T16:25:52-0400 DEBUG rpmfusion-nonfree-updates: using metadata from Sun 07 May 2023 06:10:36 PM EDT. 2023-05-12T16:25:52-0400 DEBUG User-Agent: constructed: 'libdnf (Fedora Linux 38; generic; Linux.x86_64)' 2023-05-12T16:25:53-0400 DDEBUG timer: sack setup: 10695 ms 2023-05-12T16:25:53-0400 DEBUG Completion plugin: Generating completion cache... 2023-05-12T16:25:56-0400 INFO Metadata cache created. 2023-05-12T16:25:56-0400 DDEBUG Cleaning up. 2023-05-12T16:25:56-0400 DDEBUG Plugins were unloaded. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
On Fri, May 12, 2023 at 4:52 PM Joe Zeff wrote: > > On 05/12/2023 02:36 PM, Jeffrey Walton wrote: > > Unfortunately, shutdown and reboot are not Posix commands, so they > > don't have specified behavior: > > https://pubs.opengroup.org/onlinepubs/9699919799/idx/utilities.html > > That may be, but shutdown, at least has a man page. So does the hostname command, but it no longer works, either. It would not bother me if systemd reimplemented the commands, and then replaced a command with a systemd equivalent, like hostname calling hostnamectl. But the systemd folks just broke it without a care. Jeff ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
On 05/12/2023 02:36 PM, Jeffrey Walton wrote: Unfortunately, shutdown and reboot are not Posix commands, so they don't have specified behavior: https://pubs.opengroup.org/onlinepubs/9699919799/idx/utilities.html That may be, but shutdown, at least has a man page. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
On Fri, May 12, 2023 at 3:25 AM Andras Simon wrote: > > 2023-05-04 12:58 UTC+02:00, t_pol : > > Frequently "systemctl poweroff" does NOT really "power off" the machine but > > simply halts the system. > > Does > > shutdown -h now > > work? If yes, would it be an adequate replacement? If no, I'd look > into the log files for a clue. I always thought `shutdown` was the original, and systemd was the replacement with Poettering-specified behavior :) Unfortunately, shutdown and reboot are not Posix commands, so they don't have specified behavior: https://pubs.opengroup.org/onlinepubs/9699919799/idx/utilities.html Jeff ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
On 5/12/23 10:51, Thomas Cameron wrote: I generally agree, but I was leaning towards consumer grade this time primarily for noise reasons. Enterprise class gear is just noisy as heck, and I am trying to keep the noise levels down in my home office. It's already pretty loud with the Proliants. Enterprise class switches with SFP+ modules are usually loud as heck, and expensive - you have to buy the SFP+ modules in addition to the switches. That adds up pretty quick for a home lab. Gotta keep the bride happy, you know? For short runs, you can use DAC cables; no need for separate SFP+ modules. FS.com has 2m 10G DAC cables for $14. (Just make sure that their coded properly for the hardware that you're going to use them with.) -- Google Where SkyNet meets Idiocracy ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
Thanks Roger, I'll try that. Angelo On Fri, 12 May 2023 13:02:24 -0500 Roger Heflin wrote: > you could do this and see if this works. > > stop all apps you care about and then: > sync > echo o > /proc/sysrq-trigger > > That will call the kernel directly and immediately power the machine > off, if that does not work then the kernel itself does not know how to > power it off. > > And if that does not work, then the fix would be to find a kernel > option that changes how the machine gets powered off. There used to > be options to do that, I don't know if they are still around. > > > On Fri, May 12, 2023 at 11:59 AM Patrick Mansfield via users > wrote: > > > > On Fri, May 12, 2023 at 03:00:26PM +0200, t_pol wrote: > > > On Fri, 12 May 2023 09:23:51 +0200 > > > Andras Simon wrote: > > > > > > Thanks for your answer Andras, > > > > > > To be honest that was the command I've always used > > > before "systemd" came along. > > > I don't know if it gives the same problem. > > > I'm gonna try it. > > > > > > BTW. Didn't find anything in the log files (journal). > > > > > > Ciao, > > > Angelo > > > > > > > > > > 2023-05-04 12:58 UTC+02:00, t_pol : > > > > > Hi all. > > > > > > > > > > Frequently "systemctl poweroff" does NOT really "power off" > > > > > the machine but simply halts the system. > > > > > > > > Does > > > > > > > > shutdown -h now > > > > > > > > work? If yes, would it be an adequate replacement? If no, I'd > > > > look into the log files for a clue. > > > > shutdown -h is documented as the same thing as poweroff. > > > > It could be a hardware / BIOS issue, or whatever it is in the > > kernel that sends the commands to your hardware. > > > > But if it's random ¯\_(ツ)_/¯ > > > > I have some supermicro systems I use that sometimes don't power > > off, even though AFAICT they are configured the same as others that > > power off reliably. > > > > -- Patrick > > ___ > > users mailing list -- users@lists.fedoraproject.org > > To unsubscribe send an email to users-le...@lists.fedoraproject.org > > Fedora Code of Conduct: > > https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List > > Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines > > List Archives: > > https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org > > Do not reply to spam, report it: > > https://pagure.io/fedora-infrastructure/new_issue > ___ > users mailing list -- users@lists.fedoraproject.org > To unsubscribe send an email to users-le...@lists.fedoraproject.org > Fedora Code of Conduct: > https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List > Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines > List Archives: > https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org > Do not reply to spam, report it: > https://pagure.io/fedora-infrastructure/new_issue ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
you could do this and see if this works. stop all apps you care about and then: sync echo o > /proc/sysrq-trigger That will call the kernel directly and immediately power the machine off, if that does not work then the kernel itself does not know how to power it off. And if that does not work, then the fix would be to find a kernel option that changes how the machine gets powered off. There used to be options to do that, I don't know if they are still around. On Fri, May 12, 2023 at 11:59 AM Patrick Mansfield via users wrote: > > On Fri, May 12, 2023 at 03:00:26PM +0200, t_pol wrote: > > On Fri, 12 May 2023 09:23:51 +0200 > > Andras Simon wrote: > > > > Thanks for your answer Andras, > > > > To be honest that was the command I've always used > > before "systemd" came along. > > I don't know if it gives the same problem. > > I'm gonna try it. > > > > BTW. Didn't find anything in the log files (journal). > > > > Ciao, > > Angelo > > > > > > > 2023-05-04 12:58 UTC+02:00, t_pol : > > > > Hi all. > > > > > > > > Frequently "systemctl poweroff" does NOT really "power off" the > > > > machine but simply halts the system. > > > > > > Does > > > > > > shutdown -h now > > > > > > work? If yes, would it be an adequate replacement? If no, I'd look > > > into the log files for a clue. > > shutdown -h is documented as the same thing as poweroff. > > It could be a hardware / BIOS issue, or whatever it is in the kernel that > sends the > commands to your hardware. > > But if it's random ¯\_(ツ)_/¯ > > I have some supermicro systems I use that sometimes don't power off, even > though AFAICT > they are configured the same as others that power off reliably. > > -- Patrick > ___ > users mailing list -- users@lists.fedoraproject.org > To unsubscribe send an email to users-le...@lists.fedoraproject.org > Fedora Code of Conduct: > https://docs.fedoraproject.org/en-US/project/code-of-conduct/ > List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines > List Archives: > https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org > Do not reply to spam, report it: > https://pagure.io/fedora-infrastructure/new_issue ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Firefox - detection of exiting tabs with URLs
On 5/12/23 00:54, lejeczek via users wrote: On 10/05/2023 22:17, Samuel Sieb wrote: On 5/10/23 12:47, Tim via users wrote: On Wed, 2023-05-10 at 13:04 +0200, lejeczek via users wrote: With f38 I think my Firefox is unable to detect or is ignoring tabs which are already opened in other windows with certain URL, when I open a new tab and want to go to a website/URL. I think Firefox would then say, would offer something like "switch to tab". In the past I'd noticed a behaviour that if I opened a blank tab and typed in something like facebook, let auto-complete do its thing and picked something from the drop-down list that appeared below the address bar from its history, I'd often see the browser whiz over to an existing tab which already had that site loaded. It was that or a very similar kind of situation. Look carefully at the item you're clicking. Watch for the tag that says "Switch to Tab". ___ Yes, that is exactly what we are saying, I was as the originator of this thread. Sorry, reading back now I see that was what you said, but I misunderstood it. I haven't noticed any problem, it works when I've wanted it to. It doesn't offer for any private tabs though. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: F38/Xfce - background DNF checks still going
On 5/12/23 10:08, Robert Moskowitz wrote: On 5/11/23 05:56, Patrick O'Callaghan wrote: On Thu, 2023-05-11 at 19:13 +0930, Tim via users wrote: On Wed, 2023-05-10 at 16:41 -0400, Robert Moskowitz wrote: I thought I disabled dnfdragona from doing its 3? hour check for anything new. But whenever I do a "dnf update" I see it starting with a message like: If you don't use dnf-dragora you don't need it installed. I always disabled it, then finally decided to remove it the other day. I didn't want its interrupting notices, and never used to it to install anything. I always did a "dnf update" on the command line at a time of my own choosing. Same here. Looks like I am going to have to just uninstall dhf-dragora. Even after systemctl disable dnf-makecache.timer Still got dnf update Last metadata expiration check: 0:18:17 ago on Fri 12 May 2023 12:45:30 PM EDT. Look in the logs to see what's triggering that. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: F38/Xfce - background DNF checks still going
On 5/11/23 05:56, Patrick O'Callaghan wrote: On Thu, 2023-05-11 at 19:13 +0930, Tim via users wrote: On Wed, 2023-05-10 at 16:41 -0400, Robert Moskowitz wrote: I thought I disabled dnfdragona from doing its 3? hour check for anything new. But whenever I do a "dnf update" I see it starting with a message like: If you don't use dnf-dragora you don't need it installed. I always disabled it, then finally decided to remove it the other day. I didn't want its interrupting notices, and never used to it to install anything. I always did a "dnf update" on the command line at a time of my own choosing. Same here. Looks like I am going to have to just uninstall dhf-dragora. Even after systemctl disable dnf-makecache.timer Still got dnf update Last metadata expiration check: 0:18:17 ago on Fri 12 May 2023 12:45:30 PM EDT. ARGH!!! wasting cycles and bandwidth for what? ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
On Fri, May 12, 2023 at 03:00:26PM +0200, t_pol wrote: > On Fri, 12 May 2023 09:23:51 +0200 > Andras Simon wrote: > > Thanks for your answer Andras, > > To be honest that was the command I've always used > before "systemd" came along. > I don't know if it gives the same problem. > I'm gonna try it. > > BTW. Didn't find anything in the log files (journal). > > Ciao, > Angelo > > > > 2023-05-04 12:58 UTC+02:00, t_pol : > > > Hi all. > > > > > > Frequently "systemctl poweroff" does NOT really "power off" the > > > machine but simply halts the system. > > > > Does > > > > shutdown -h now > > > > work? If yes, would it be an adequate replacement? If no, I'd look > > into the log files for a clue. shutdown -h is documented as the same thing as poweroff. It could be a hardware / BIOS issue, or whatever it is in the kernel that sends the commands to your hardware. But if it's random ¯\_(ツ)_/¯ I have some supermicro systems I use that sometimes don't power off, even though AFAICT they are configured the same as others that power off reliably. -- Patrick ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
On 5/12/23 10:39, Roger Heflin wrote: yes, that would give way more speed if those desktops need that kind of speed. "Need" is relative. ;-) But I use one of the servers as a hypervisor for testing stuff, and the other is a file server. I'm mostly thinking about the file server, that's where I export /home and then mount it on my desktop. I'd like that to be faster. I have had bad enough luck with consumer grade SATA/SAS cards and nic cards that the used enterprise grade of either is a lot more likely to work reliably. I generally do the same thing, my enterprise gear is all used and I have a vendor in Houston I have had really good luck with (ServerMonkey). I won't touch a consumer brand of that sort of stuff (unless it comes with the MB) simply because when I need a card I can get a used enterprise for about the same price and know that it will work. I generally agree, but I was leaning towards consumer grade this time primarily for noise reasons. Enterprise class gear is just noisy as heck, and I am trying to keep the noise levels down in my home office. It's already pretty loud with the Proliants. Enterprise class switches with SFP+ modules are usually loud as heck, and expensive - you have to buy the SFP+ modules in addition to the switches. That adds up pretty quick for a home lab. Gotta keep the bride happy, you know? Thomas ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
yes, that would give way more speed if those desktops need that kind of speed. I have had bad enough luck with consumer grade SATA/SAS cards and nic cards that the used enterprise grade of either is a lot more likely to work reliably. I won't touch a consumer brand of that sort of stuff (unless it comes with the MB) simply because when I need a card I can get a used enterprise for about the same price and know that it will work. On Fri, May 12, 2023 at 10:35 AM Thomas Cameron wrote: > > On 5/12/23 10:14, Roger Heflin wrote: > > Given that hardware, buy something like this instead. > > > > used, but better class of card. > > > > https://www.newegg.com/intel-e10g42bt/p/N82E16833106075?Description=10gbit%20card_re=10gbit_card-_-33-106-075-_-Product=true > > > > or something similar from the used sellers. There seem to be a decent > > variety of cards under $100. > > > > Manufacturers that are good are HPE/DELL/Intel/Broadcom/IBM/Lenovo. > > But you do need to do a bit of research on the given cards to see what > > the real chipset is. Avoid Emulex/Be2net variants they have "issues". > > Intel/Broadcom/Mellanox based cards are good. > > This is awesome, thank you very much! I'm still trying to get up to > speed on 10G networking, I genuinely appreciate this advice. > > With that two port card, I could get three and just run direct cables > from my desktop to the servers, then a cable between the servers. No > switch required, and it would be fast as heck. > > Thanks! > Thomas > ___ > users mailing list -- users@lists.fedoraproject.org > To unsubscribe send an email to users-le...@lists.fedoraproject.org > Fedora Code of Conduct: > https://docs.fedoraproject.org/en-US/project/code-of-conduct/ > List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines > List Archives: > https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org > Do not reply to spam, report it: > https://pagure.io/fedora-infrastructure/new_issue ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
On 5/12/23 10:19, Roger Heflin wrote: And also given your hw, limit your interrupt count to the number of cores on a single socket, the interrupts will typically only happen on the local socket the given pci bus is attached to, and you would need to determine if different cards have their PCI buses connected to different sockets. Will do, thanks. Which model of proliant is that? Note I have used/debugged/optimized dl360/dl380/dl560/dl580/dl980 gen7/8/9/10/11(testing). Currently Gen 8 DL380 with 12 3.5" disks. I'm about to upgrade to a couple of Gen 9 servers and probably retire/sell the Gen 8s. Thomas ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
On 5/12/23 10:14, Roger Heflin wrote: Given that hardware, buy something like this instead. used, but better class of card. https://www.newegg.com/intel-e10g42bt/p/N82E16833106075?Description=10gbit%20card_re=10gbit_card-_-33-106-075-_-Product=true or something similar from the used sellers. There seem to be a decent variety of cards under $100. Manufacturers that are good are HPE/DELL/Intel/Broadcom/IBM/Lenovo. But you do need to do a bit of research on the given cards to see what the real chipset is. Avoid Emulex/Be2net variants they have "issues". Intel/Broadcom/Mellanox based cards are good. This is awesome, thank you very much! I'm still trying to get up to speed on 10G networking, I genuinely appreciate this advice. With that two port card, I could get three and just run direct cables from my desktop to the servers, then a cable between the servers. No switch required, and it would be fast as heck. Thanks! Thomas ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
And also given your hw, limit your interrupt count to the number of cores on a single socket, the interrupts will typically only happen on the local socket the given pci bus is attached to, and you would need to determine if different cards have their PCI buses connected to different sockets. Which model of proliant is that? Note I have used/debugged/optimized dl360/dl380/dl560/dl580/dl980 gen7/8/9/10/11(testing). On Fri, May 12, 2023 at 10:14 AM Roger Heflin wrote: > > Given that hardware, buy something like this instead. > > used, but better class of card. > > https://www.newegg.com/intel-e10g42bt/p/N82E16833106075?Description=10gbit%20card_re=10gbit_card-_-33-106-075-_-Product=true > > or something similar from the used sellers. There seem to be a decent > variety of cards under $100. > > Manufacturers that are good are HPE/DELL/Intel/Broadcom/IBM/Lenovo. > But you do need to do a bit of research on the given cards to see what > the real chipset is. Avoid Emulex/Be2net variants they have "issues". > Intel/Broadcom/Mellanox based cards are good. > > On Fri, May 12, 2023 at 10:06 AM Thomas Cameron > wrote: > > > > On 5/12/23 09:41, Roger Heflin wrote: > > > I have work experience with Intel 10Gbit, Older Emulex 10GB (be[23]net > > > driver), current Broadcom 10/25G and Mellanox 10/25G. > > > > > > Note that for it to be useful you*MUST* have multiple interrupts. > > > 1Gbit interfaces used to hit a limit at around 50Mbit (the cpu was not > > > fast enough to do the required data moves on a single core to go much > > > faster). Given the cpus are faster now the limit is probably up to > > > 1Gbit/sec/interrupt. > > > > > > I have had to set nic adaptors to use a layer-3 data -> interrupt so > > > that you can get higher rates (ethtool setting) only available on some > > > nic cards. > > > > > > I have also had to set the interrupt number (ethtool setting), note > > > that it is rather pointless to have the interrupt count higher than > > > the number of real cores. And likely you want the interrupt count for > > > the disk controllers being used + the nic interrupts <= the number of > > > cores. > > > > > > At home I question the value of it. You might even simply test your > > > nfs home setup and see if you can even get close to Gbit speeds, > > > unless you have a newer machine and a good underlying disk setup you > > > probably aren't going to get there. And you also need a decent Gbit > > > card, but if you do not have a decent gbit card, you can buy a used > > > enterprise grade dual gbit card for around $25 (branded Dell or > > > branded HPE). > > > > > > With enterprise grade hw and good disk and other parts one can hit > > > 115-125MBytes/sec. > > > > > > With a new machine at home and a 7-disk raid-6 setup with a good gbit > > > nic card I can get 115-125MB/sec when data is in cache, but not coming > > > directly off of disk, unless the disk is an ssd and reading large > > > files. My prior machine could not hit that rate but was 10 year old > > > hw. > > > > > > To use a 10Gbit interface you will have to have multiple machines > > > doing large file sequential io (assuming they are wireless or gbit > > > interfaces) at the same time. > > > > > > I doubt you are going to gain enough (or possibly any) to make it > > > worth the cost or the trouble to get it working. > > > > > > Install sar and configure it to sample at 1minute and sar -n DEV will > > > show you your network rates, if you aren't currently sustaining > > > 115MByte/sec every so often then 10gbit is not going to do anything > > > for you. > > > > My home office has a couple of HP Proliant servers with big (12 SAS > > drives) RAID arrays and dual CPU E5-2697 v2 processors with 24 cores > > each (48 cores total in the machine). I can sustain 1.6Gb/sec reads and > > writes, tested via dd with oflag=dsync and also with fio. > > > > So with consumer grade NICs on the Proliants and my workstation > > (https://www.amazon.com/TP-Link-TX401-Ethernet-Supports-Including/dp/B08D71PVXG/), > > I suspect it will give me better performance than the 1Gb ethernet I'm > > currently using. Will it be as amazing as high end enterprise gear? Most > > likely not. Will it be better than 1Gb ethernet I'm currently using? I > > suspect so. > > > > I appreciate the advice, I'll definitely play around with the settings > > to make sure I'm using enough cores. > > > > Thomas > > ___ > > users mailing list -- users@lists.fedoraproject.org > > To unsubscribe send an email to users-le...@lists.fedoraproject.org > > Fedora Code of Conduct: > > https://docs.fedoraproject.org/en-US/project/code-of-conduct/ > > List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines > > List Archives: > > https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org > > Do not reply to spam, report it: > > https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
Once upon a time, Roger Heflin said: > To use a 10Gbit interface you will have to have multiple machines > doing large file sequential io (assuming they are wireless or gbit > interfaces) at the same time. I disagree. Both my home "server" and desktop are regular desktop motherboards (well, "gaming" models because those usually have more ports), with Ryzen CPUs (nothing particularly fancy). The server has 3 NAS-type SATA drives in Linux md RAID5, with an NVMe read cache on the LVM pool. I just picked a random large file that wasn't in the cache and did a dd over the NFS and got 199 MB/s - hot cache it was 1.2 GB/s. This is with zero effort at tuning the network interfaces (managing interrupts or any ethtool-type settings), or even really doing much to try to improve NAS speed (like using more and/or faster drives). Modern systems can easily surpass what a 1 gigabit NIC can do. -- Chris Adams ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
Given that hardware, buy something like this instead. used, but better class of card. https://www.newegg.com/intel-e10g42bt/p/N82E16833106075?Description=10gbit%20card_re=10gbit_card-_-33-106-075-_-Product=true or something similar from the used sellers. There seem to be a decent variety of cards under $100. Manufacturers that are good are HPE/DELL/Intel/Broadcom/IBM/Lenovo. But you do need to do a bit of research on the given cards to see what the real chipset is. Avoid Emulex/Be2net variants they have "issues". Intel/Broadcom/Mellanox based cards are good. On Fri, May 12, 2023 at 10:06 AM Thomas Cameron wrote: > > On 5/12/23 09:41, Roger Heflin wrote: > > I have work experience with Intel 10Gbit, Older Emulex 10GB (be[23]net > > driver), current Broadcom 10/25G and Mellanox 10/25G. > > > > Note that for it to be useful you*MUST* have multiple interrupts. > > 1Gbit interfaces used to hit a limit at around 50Mbit (the cpu was not > > fast enough to do the required data moves on a single core to go much > > faster). Given the cpus are faster now the limit is probably up to > > 1Gbit/sec/interrupt. > > > > I have had to set nic adaptors to use a layer-3 data -> interrupt so > > that you can get higher rates (ethtool setting) only available on some > > nic cards. > > > > I have also had to set the interrupt number (ethtool setting), note > > that it is rather pointless to have the interrupt count higher than > > the number of real cores. And likely you want the interrupt count for > > the disk controllers being used + the nic interrupts <= the number of > > cores. > > > > At home I question the value of it. You might even simply test your > > nfs home setup and see if you can even get close to Gbit speeds, > > unless you have a newer machine and a good underlying disk setup you > > probably aren't going to get there. And you also need a decent Gbit > > card, but if you do not have a decent gbit card, you can buy a used > > enterprise grade dual gbit card for around $25 (branded Dell or > > branded HPE). > > > > With enterprise grade hw and good disk and other parts one can hit > > 115-125MBytes/sec. > > > > With a new machine at home and a 7-disk raid-6 setup with a good gbit > > nic card I can get 115-125MB/sec when data is in cache, but not coming > > directly off of disk, unless the disk is an ssd and reading large > > files. My prior machine could not hit that rate but was 10 year old > > hw. > > > > To use a 10Gbit interface you will have to have multiple machines > > doing large file sequential io (assuming they are wireless or gbit > > interfaces) at the same time. > > > > I doubt you are going to gain enough (or possibly any) to make it > > worth the cost or the trouble to get it working. > > > > Install sar and configure it to sample at 1minute and sar -n DEV will > > show you your network rates, if you aren't currently sustaining > > 115MByte/sec every so often then 10gbit is not going to do anything > > for you. > > My home office has a couple of HP Proliant servers with big (12 SAS > drives) RAID arrays and dual CPU E5-2697 v2 processors with 24 cores > each (48 cores total in the machine). I can sustain 1.6Gb/sec reads and > writes, tested via dd with oflag=dsync and also with fio. > > So with consumer grade NICs on the Proliants and my workstation > (https://www.amazon.com/TP-Link-TX401-Ethernet-Supports-Including/dp/B08D71PVXG/), > I suspect it will give me better performance than the 1Gb ethernet I'm > currently using. Will it be as amazing as high end enterprise gear? Most > likely not. Will it be better than 1Gb ethernet I'm currently using? I > suspect so. > > I appreciate the advice, I'll definitely play around with the settings > to make sure I'm using enough cores. > > Thomas > ___ > users mailing list -- users@lists.fedoraproject.org > To unsubscribe send an email to users-le...@lists.fedoraproject.org > Fedora Code of Conduct: > https://docs.fedoraproject.org/en-US/project/code-of-conduct/ > List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines > List Archives: > https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org > Do not reply to spam, report it: > https://pagure.io/fedora-infrastructure/new_issue ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
On 5/12/23 10:03, Chris Adams wrote: Once upon a time, Thomas Cameron said: Do you just use a copper SFP+ module like https://www.ebay.com/itm/164322691847 in the Microtik? I'd love to know what you use. I used QFPTEK modules from Amazon, but any should do. The Mikrotik does run a little hot with them (it's passively cooled), but I have it vertically mounted and it seems fine. I was leaning towards this switch: https://www.amazon.com/TP-Link-TL-SX105-Wall-Mount-Protection-Auto-Negotiation/dp/B09CYNHL4S. It looks super simple, it already has the copper interfaces so I don't need SFP+ modules. I went with the Mikrotik because I wanted a managed switch so I could have VLANs. My preference is to use plain old cat7 ethernet, like 10 foot cables or so: https://www.amazon.com/AmazonBasics-High-Speed-Gigabit-Ethernet-Internet/dp/B07ZTQY9DD/ I went with some CableCreation "ultra thin" cat6A cables (makes for easier cable routing and bundling). This is gold, thank you so much! I've only used InfiniBand in my home network (got some really cheap 10G IB stuff) and it was really fast. I loved it. But I sold that stuff when I changed jobs, and I have a bad itch to get back to faster, but more modern networking. Many thanks for your advice! Thomas ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
On 5/12/23 09:41, Roger Heflin wrote: I have work experience with Intel 10Gbit, Older Emulex 10GB (be[23]net driver), current Broadcom 10/25G and Mellanox 10/25G. Note that for it to be useful you*MUST* have multiple interrupts. 1Gbit interfaces used to hit a limit at around 50Mbit (the cpu was not fast enough to do the required data moves on a single core to go much faster). Given the cpus are faster now the limit is probably up to 1Gbit/sec/interrupt. I have had to set nic adaptors to use a layer-3 data -> interrupt so that you can get higher rates (ethtool setting) only available on some nic cards. I have also had to set the interrupt number (ethtool setting), note that it is rather pointless to have the interrupt count higher than the number of real cores. And likely you want the interrupt count for the disk controllers being used + the nic interrupts <= the number of cores. At home I question the value of it. You might even simply test your nfs home setup and see if you can even get close to Gbit speeds, unless you have a newer machine and a good underlying disk setup you probably aren't going to get there. And you also need a decent Gbit card, but if you do not have a decent gbit card, you can buy a used enterprise grade dual gbit card for around $25 (branded Dell or branded HPE). With enterprise grade hw and good disk and other parts one can hit 115-125MBytes/sec. With a new machine at home and a 7-disk raid-6 setup with a good gbit nic card I can get 115-125MB/sec when data is in cache, but not coming directly off of disk, unless the disk is an ssd and reading large files. My prior machine could not hit that rate but was 10 year old hw. To use a 10Gbit interface you will have to have multiple machines doing large file sequential io (assuming they are wireless or gbit interfaces) at the same time. I doubt you are going to gain enough (or possibly any) to make it worth the cost or the trouble to get it working. Install sar and configure it to sample at 1minute and sar -n DEV will show you your network rates, if you aren't currently sustaining 115MByte/sec every so often then 10gbit is not going to do anything for you. My home office has a couple of HP Proliant servers with big (12 SAS drives) RAID arrays and dual CPU E5-2697 v2 processors with 24 cores each (48 cores total in the machine). I can sustain 1.6Gb/sec reads and writes, tested via dd with oflag=dsync and also with fio. So with consumer grade NICs on the Proliants and my workstation (https://www.amazon.com/TP-Link-TX401-Ethernet-Supports-Including/dp/B08D71PVXG/), I suspect it will give me better performance than the 1Gb ethernet I'm currently using. Will it be as amazing as high end enterprise gear? Most likely not. Will it be better than 1Gb ethernet I'm currently using? I suspect so. I appreciate the advice, I'll definitely play around with the settings to make sure I'm using enough cores. Thomas ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
Once upon a time, Thomas Cameron said: > Do you just use a copper SFP+ module like > https://www.ebay.com/itm/164322691847 in the Microtik? I'd love to > know what you use. I used QFPTEK modules from Amazon, but any should do. The Mikrotik does run a little hot with them (it's passively cooled), but I have it vertically mounted and it seems fine. > I was leaning towards this switch: > https://www.amazon.com/TP-Link-TL-SX105-Wall-Mount-Protection-Auto-Negotiation/dp/B09CYNHL4S. > It looks super simple, it already has the copper interfaces so I > don't need SFP+ modules. I went with the Mikrotik because I wanted a managed switch so I could have VLANs. > My preference is to use plain old cat7 ethernet, like 10 foot cables > or so: > https://www.amazon.com/AmazonBasics-High-Speed-Gigabit-Ethernet-Internet/dp/B07ZTQY9DD/ I went with some CableCreation "ultra thin" cat6A cables (makes for easier cable routing and bundling). -- Chris Adams ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
I have work experience with Intel 10Gbit, Older Emulex 10GB (be[23]net driver), current Broadcom 10/25G and Mellanox 10/25G. Note that for it to be useful you *MUST* have multiple interrupts. 1Gbit interfaces used to hit a limit at around 50Mbit (the cpu was not fast enough to do the required data moves on a single core to go much faster). Given the cpus are faster now the limit is probably up to 1Gbit/sec/interrupt. I have had to set nic adaptors to use a layer-3 data -> interrupt so that you can get higher rates (ethtool setting) only available on some nic cards. I have also had to set the interrupt number (ethtool setting), note that it is rather pointless to have the interrupt count higher than the number of real cores. And likely you want the interrupt count for the disk controllers being used + the nic interrupts <= the number of cores. At home I question the value of it. You might even simply test your nfs home setup and see if you can even get close to Gbit speeds, unless you have a newer machine and a good underlying disk setup you probably aren't going to get there. And you also need a decent Gbit card, but if you do not have a decent gbit card, you can buy a used enterprise grade dual gbit card for around $25 (branded Dell or branded HPE). With enterprise grade hw and good disk and other parts one can hit 115-125MBytes/sec. With a new machine at home and a 7-disk raid-6 setup with a good gbit nic card I can get 115-125MB/sec when data is in cache, but not coming directly off of disk, unless the disk is an ssd and reading large files. My prior machine could not hit that rate but was 10 year old hw. To use a 10Gbit interface you will have to have multiple machines doing large file sequential io (assuming they are wireless or gbit interfaces) at the same time. I doubt you are going to gain enough (or possibly any) to make it worth the cost or the trouble to get it working. Install sar and configure it to sample at 1minute and sar -n DEV will show you your network rates, if you aren't currently sustaining 115MByte/sec every so often then 10gbit is not going to do anything for you. On Fri, May 12, 2023 at 8:36 AM Thomas Cameron wrote: > > Howdy, all - > > I use an NFS server export to mount my /home directory on my desktop. > I've got the itch to go to 10Gb ethernet, but I am reading that the > tp-link tx401 has a problem with bridging, and I use bridging for KVM > virtual machines on my desktop. I *think* that you can just disable > using the command "ethtool -K lro off," but I wondered if anyone > had any experience with NICs that work with bridging out of the box. > > Anyone got any experience with 10Gb ethernet cards? Good? Bad? Hassles? > > Thomas > ___ > users mailing list -- users@lists.fedoraproject.org > To unsubscribe send an email to users-le...@lists.fedoraproject.org > Fedora Code of Conduct: > https://docs.fedoraproject.org/en-US/project/code-of-conduct/ > List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines > List Archives: > https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org > Do not reply to spam, report it: > https://pagure.io/fedora-infrastructure/new_issue ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
On 5/12/23 08:53, Chris Adams wrote: I have a couple of these, one in my home server (which includes NFS) and one in my primary desktop, connected through a Mikrotik CRS305 switch, and the setup works fine. I am using bridging on both systems (for VMs), and that works fine as well. Checking my interfaces, it looks like LRO is already disabled (I'm guessing by the "atlantic" kernel driver, as I haven't set any ethtool options). I have experienced a couple of issues, both related to putting my desktop to sleep when I'm not using it: - Every once in a while, when I resume, the network is dead. There's a kernel oops (that's kind of vague) and I have to reboot. This isn't a huge problem, because it only seems to happen after a bunch of suspend/resume cycles, and I typically reboot for updates more often than it happens. - I also run jumbo frames, and after a suspend/resume cycle, the MTU on the NIC resets to 1500 (while the bridge interface stays jumbo). This breaks communication. I don't know if the driver is expected to restore the MTU and isn't, but NetworkManager also doesn't seem to handle bridge+suspend/resume right; NICs show "connected (externally)" in nmcli after a suspend/resume, like NM loses management of them. I've just hacked around this by adding a dispatcher script to reset the MTU. Do you just use a copper SFP+ module like https://www.ebay.com/itm/164322691847 in the Microtik? I'd love to know what you use. I was leaning towards this switch: https://www.amazon.com/TP-Link-TL-SX105-Wall-Mount-Protection-Auto-Negotiation/dp/B09CYNHL4S. It looks super simple, it already has the copper interfaces so I don't need SFP+ modules. My preference is to use plain old cat7 ethernet, like 10 foot cables or so: https://www.amazon.com/AmazonBasics-High-Speed-Gigabit-Ethernet-Internet/dp/B07ZTQY9DD/ Thomas ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Firefox - detection of exiting tabs with URLs
On Fri, 2023-05-12 at 11:38 +0100, Patrick O'Callaghan wrote: > Looks like you should report this upstream. I doubt it's a Fedora bug. > > poc POC, you didn't trim your quotes! ;-) I had to wait a long time to get that joke in. -- NB: All unexpected mail to my mailbox is automatically deleted. I will only get to see the messages that are posted to the list. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Experience with 10Gb ethernet adapters?
Once upon a time, Thomas Cameron said: > I use an NFS server export to mount my /home directory on my > desktop. I've got the itch to go to 10Gb ethernet, but I am reading > that the tp-link tx401 has a problem with bridging, and I use > bridging for KVM virtual machines on my desktop. I *think* that you > can just disable using the command "ethtool -K lro off," but > I wondered if anyone had any experience with NICs that work with > bridging out of the box. I have a couple of these, one in my home server (which includes NFS) and one in my primary desktop, connected through a Mikrotik CRS305 switch, and the setup works fine. I am using bridging on both systems (for VMs), and that works fine as well. Checking my interfaces, it looks like LRO is already disabled (I'm guessing by the "atlantic" kernel driver, as I haven't set any ethtool options). I have experienced a couple of issues, both related to putting my desktop to sleep when I'm not using it: - Every once in a while, when I resume, the network is dead. There's a kernel oops (that's kind of vague) and I have to reboot. This isn't a huge problem, because it only seems to happen after a bunch of suspend/resume cycles, and I typically reboot for updates more often than it happens. - I also run jumbo frames, and after a suspend/resume cycle, the MTU on the NIC resets to 1500 (while the bridge interface stays jumbo). This breaks communication. I don't know if the driver is expected to restore the MTU and isn't, but NetworkManager also doesn't seem to handle bridge+suspend/resume right; NICs show "connected (externally)" in nmcli after a suspend/resume, like NM loses management of them. I've just hacked around this by adding a dispatcher script to reset the MTU. -- Chris Adams ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Experience with 10Gb ethernet adapters?
Howdy, all - I use an NFS server export to mount my /home directory on my desktop. I've got the itch to go to 10Gb ethernet, but I am reading that the tp-link tx401 has a problem with bridging, and I use bridging for KVM virtual machines on my desktop. I *think* that you can just disable using the command "ethtool -K lro off," but I wondered if anyone had any experience with NICs that work with bridging out of the box. Anyone got any experience with 10Gb ethernet cards? Good? Bad? Hassles? Thomas ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
On Fri, 12 May 2023 09:23:51 +0200 Andras Simon wrote: Thanks for your answer Andras, To be honest that was the command I've always used before "systemd" came along. I don't know if it gives the same problem. I'm gonna try it. BTW. Didn't find anything in the log files (journal). Ciao, Angelo > 2023-05-04 12:58 UTC+02:00, t_pol : > > Hi all. > > > > Frequently "systemctl poweroff" does NOT really "power off" the > > machine but simply halts the system. > > Does > > shutdown -h now > > work? If yes, would it be an adequate replacement? If no, I'd look > into the log files for a clue. > ___ > users mailing list -- users@lists.fedoraproject.org > To unsubscribe send an email to users-le...@lists.fedoraproject.org > Fedora Code of Conduct: > https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List > Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines > List Archives: > https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org > Do not reply to spam, report it: > https://pagure.io/fedora-infrastructure/new_issue ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Firefox - detection of exiting tabs with URLs
On Fri, 2023-05-12 at 09:54 +0200, lejeczek via users wrote: > > > On 10/05/2023 22:17, Samuel Sieb wrote: > > On 5/10/23 12:47, Tim via users wrote: > > > On Wed, 2023-05-10 at 13:04 +0200, lejeczek via users wrote: > > > > With f38 I think my Firefox is unable to detect or is > > > > ignoring tabs > > > > which are already opened in other windows with certain > > > > URL, when I > > > > open a new tab and want to go to a website/URL. > > > > I think Firefox would then say, would offer something > > > > like "switch > > > > to tab". > > > > > > In the past I'd noticed a behaviour that if I opened a > > > blank tab and > > > typed in something like facebook, let auto-complete do > > > its thing and > > > picked something from the drop-down list that appeared > > > below the > > > address bar from its history, I'd often see the browser > > > whiz over to an > > > existing tab which already had that site loaded. It was > > > that or a very > > > similar kind of situation. > > > > Look carefully at the item you're clicking. Watch for the > > tag that says "Switch to Tab". > > ___ > > > Yes, that is exactly what we are saying, I was as the > originator of this thread. > I can only add - to what I said which was "this > functionality is completely gone" - that now after a longer > look I see "Swtich to Tab" for some site/urls, sometimes! > but not the others. > This would need to be thoroughly tested (probably something > only devel could do) to say, if there is pattern to it at > all or all this is purely random. > Eg. I've had a window with a few tabs - my Firefox is set to > restore last sessions - _none_ of those tabs (sites/urls > were fully loaded after restoration I made sure of) was > detected as "Switch to Tab" > I closed that window, opened a new one (made sure that no > other auto-started/restored window/tab had those sites in > question/example opened) and each site, one by one > loaded/opened, then tested in already existing window and > only ! then that window picked up, detected and offered > "Switch to Tab" to those sites. > So, purely from consumer perspective I can say this > "feature" is bit flaky. Looks like you should report this upstream. I doubt it's a Fedora bug. poc ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Firefox - detection of exiting tabs with URLs
On 10/05/2023 22:17, Samuel Sieb wrote: On 5/10/23 12:47, Tim via users wrote: On Wed, 2023-05-10 at 13:04 +0200, lejeczek via users wrote: With f38 I think my Firefox is unable to detect or is ignoring tabs which are already opened in other windows with certain URL, when I open a new tab and want to go to a website/URL. I think Firefox would then say, would offer something like "switch to tab". In the past I'd noticed a behaviour that if I opened a blank tab and typed in something like facebook, let auto-complete do its thing and picked something from the drop-down list that appeared below the address bar from its history, I'd often see the browser whiz over to an existing tab which already had that site loaded. It was that or a very similar kind of situation. Look carefully at the item you're clicking. Watch for the tag that says "Switch to Tab". ___ Yes, that is exactly what we are saying, I was as the originator of this thread. I can only add - to what I said which was "this functionality is completely gone" - that now after a longer look I see "Swtich to Tab" for some site/urls, sometimes! but not the others. This would need to be thoroughly tested (probably something only devel could do) to say, if there is pattern to it at all or all this is purely random. Eg. I've had a window with a few tabs - my Firefox is set to restore last sessions - _none_ of those tabs (sites/urls were fully loaded after restoration I made sure of) was detected as "Switch to Tab" I closed that window, opened a new one (made sure that no other auto-started/restored window/tab had those sites in question/example opened) and each site, one by one loaded/opened, then tested in already existing window and only ! then that window picked up, detected and offered "Switch to Tab" to those sites. So, purely from consumer perspective I can say this "feature" is bit flaky. many thanks, L. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Re: Poweroff on Fedora 37
2023-05-04 12:58 UTC+02:00, t_pol : > Hi all. > > Frequently "systemctl poweroff" does NOT really "power off" the machine but > simply halts the system. Does shutdown -h now work? If yes, would it be an adequate replacement? If no, I'd look into the log files for a clue. ___ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue