1516413962 Q * Aiken Remote host closed the connection 1516414150 J * Aiken ~Aiken@2001:44b8:2168:1000:b26e:bfff:fe2a:b951 1516415524 Q * obeardly Quit: Leaving 1516416369 Q * Aiken Remote host closed the connection 1516416593 J * Aiken ~Aiken@2001:44b8:2168:1000:b26e:bfff:fe2a:b951 1516418498 J * ksn ~ksn@00019501.user.oftc.net 1516418590 Q * ksn 1516421955 J * obeardly ~obeardly@50-240-103-34-static.hfc.comcastbusiness.net 1516423650 Q * obeardly Quit: Leaving 1516426659 N * Bertl_oO Bertl_zZ 1516435626 Q * romster Remote host closed the connection 1516435654 J * romster ~romster@158.140.215.184 1516435937 Q * romster Remote host closed the connection 1516435960 J * romster ~romster@158.140.215.184 1516436033 Q * romster Remote host closed the connection 1516436059 J * romster ~romster@158.140.215.184 1516438723 J * fstd_ ~fstd@xdsl-78-35-66-189.netcologne.de 1516439173 Q * fstd Ping timeout: 480 seconds 1516439173 N * fstd_ fstd 1516439852 M * Ghislain1 with 4.4 but the same apply to 4.9 1516439995 M * Le_Coyote Ghislain1: does that solve the issue in free and htop? 1516440008 M * Ghislain1 yes Bertl told me not to do that, division is a risk in the kernel 1516440022 M * Ghislain1 i dont know how to do that without it 1516440617 M * Ghislain1 only solution i see is to test for mem_unit == 1 or page_SIZE anyother case would fail 1516440887 M * yang_ Hello, I am still trying to restore that drive from fsck-errors. Those were somehow sorted by "Guy" but then the grub error appeared, and I was told to change the partition table from GPT back to MBR because its supposed to be more optimized with older drives...So maybe at that point I could write in the Grub again ? Could I get some assistance with that, I don't want to do a mistake. 1516441435 M * Ghislain1 sorry out of my knowledge base i would do advise that are not safe to follow :) 1516441952 M * Ghislain1 ok i put a version of my fix that does test mem_unit for 1 and if not it default to PAGE_SIZE 1516441979 M * Ghislain1 search in the kernel seems to indicate that only those 2 case are possible 1516442045 M * Aiken Ghislain1, where is the problem you are seeing showing up? 1516442050 M * Ghislain1 Bertl_zZ: so what is not yet working on 4.4 and 4.9 ? virt_mem seems to be potentialy fixed, so vitr_load ? any other ? 1516442062 M * Ghislain1 Aiken: virt_mem was not working 1516442082 M * Ghislain1 reporting used memory was really wrong 1516442152 M * Ghislain1 seems the default mem_unit of the sysinfo struct is not PAGE_SIZE but 1 on 4.4+, i thing 32bits are PAGE_SIZE, 64bits are 1 (because the cpu counters are 64bits so why not put the value instead of something needing an operation 1516442200 M * Ghislain1 i sent my patch to the list if anyone want to have alook at it 1516442285 M * Ghislain1 i have done a little C program that return printf ("mem_unit   : %d bytes\n", pointer_to_sysinfo.mem_unit), this is when i saw there was a change here 1516443823 Q * Guy- Ping timeout: 480 seconds 1516443848 M * Aiken with 4.9 I see mem_unit = 1. Top is showing expected values for me on the host, in the guest, on a non vserver machine 1516443983 M * Ghislain1 guest should show you pages instead of usage in bytes 1516444048 M * Aiken # free -h 1516444048 M * Aiken total used free shared buff/cache available 1516444048 M * Aiken Mem: 7.8G 54M 7.5G 708K 231M 7.6G 1516444048 M * Aiken Swap: 9.0G 0B 9.0G 1516444054 M * Aiken this are correct 1516444116 M * Ghislain1 you guest use 54mb ? 1516444176 M * Ghislain1 which kernel and vserver version do you use ? 1516444227 M * Aiken same ram as the host free is showing. I am used to quests showing the same ram usage as the host 1516444231 M * Aiken 4.9.76 1516444247 M * Aiken 4.9.76-vs2.3.9.5 1516444261 M * Ghislain1 well if you do not limit ram in guest and dont have virt_mem then the free show the host value not the guest values 1516444285 M * Ghislain1 virt_mem show usage of the guest, not the host when you are in the guest 1516444339 M * Ghislain1 if you put VIRT_MEM in cflags, like virt_load etc.. 1516445115 N * guerby_ guerby 1516446406 M * Aiken total used free shared buff/cache available 1516446406 M * Aiken Mem: 256M 4K 255M 708K 11M 255M 1516446406 M * Aiken Swap: 128M 0B 128M 1516446422 M * Aiken same kernel 1516446594 J * Guy- ~korn@elan.rulez.org 1516447252 M * Aiken downgrading to 4.4.111, will take at least 1/2 hour till a bootable kernel 1516447980 N * Bertl_zZ Bertl 1516447983 M * Bertl morning folks! 1516448033 M * Aiken good evening 1516448077 M * Bertl Ghislain1: the new approach looks better, but I'm a little confused by the condition 1516448154 M * Bertl when val->mem_unit = 1 (i.e. we expect bytes, no?) then you shift the values to the right effectively dividing them by the page size? 1516448219 M * Bertl isn't that the wrong way around? 1516448604 M * yang_ Bertl: Did you receive my paypal payment ? 1516448625 M * Bertl yes, I did! thanks! appreciated! 1516448637 M * yang_ okay 1516448657 M * yang_ welcome 1516448695 M * Bertl as far as I understood, thanks also go to Guy- here 1516448707 M * yang_ Yes, he did all the work 1516448755 M * yang_ Now I am just doing a dd copy of the old drive partitions, remotelly 1516448767 M * yang_ and later I was suggested to replace the grub with 1516448777 M * yang_ extlinux bootloader 1516448788 M * Bertl that is a good approach, btw, you can use lzop to compress such a dump on-the-fly 1516448791 M * yang_ so that might work, if it doesnt no harm done, as I will have backup 1516448828 M * yang_ i used "ssh | dd if=/dev/sda3 of=/dev/stdout bs=1M | lz4 | pv > backup.img.lz4 1516448916 M * Bertl that won't work, but I presume you are missing some parts (at least quotes or so) 1516448924 M * yang_ well 1516448927 M * yang_ ssh part is 1516448929 M * yang_ different 1516448952 M * yang_ ssh dd if=/dev/sda3 of=/dev/stdout bs=1M | lz4 | pv > backup.img.lz4 1516448992 M * yang_ 1516449013 M * Bertl with everything from dd to pv on the remote host, I presume 1516449039 M * Bertl but the redirection on the local (your backup) system, yes? 1516449063 M * yang_ yes, remote host 1516449087 M * Bertl okay, looks fine, although you can skip the of=/dev/stdout, because that is the default 1516449148 M * Bertl a block size of 128k might also give a slightly better performance (not much) and if the ssh/sshd allows it, you could use arc4 as crypto instead of the default, which speeds up things a little too 1516449891 J * Ghislain ~ghislain@81.56.195.31 1516450173 Q * Ghislain1 Ping timeout: 480 seconds 1516450413 M * yang_ ok bertl, thanks for hints 1516450542 M * Guy- yang_: it would be better to compress on the remote end, before sending, but this should also work 1516450614 M * yang_ ok Guy- 1516450703 J * Ghislain1 ~ghislain@adsl1.aqueos.com 1516450973 Q * Ghislain Ping timeout: 480 seconds 1516451578 M * Ghislain1 Bertl: what confuse you ? if mem_unit is byte with convert pages to bytes 1516451597 M * Ghislain1 if it is other ( PAGE_SIZE) then we let it as we are in pages allready 1516451613 M * Ghislain1 ah that is the shift ok 1516451708 M * Ghislain1 i dont know how this works so i had 50% chance of being ok 1516451822 M * Ghislain1 this is what happen when you let an system admin do C like he do english 1516454770 Q * Aiken Remote host closed the connection 1516456824 M * Ghislain1 Bertl: well the more i try to fix it the more it fail :( my first on worked but was wrong for the kernel 1516456838 M * Ghislain1 my oether try all fails in various way 1516457932 Q * Ghislain1 Ping timeout: 480 seconds 1516458239 J * Ghislain ~ghislain@adsl1.aqueos.com 1516458898 M * Ghislain well the lmast try seems pretty ok 1516458925 M * Ghislain Bertl: must go but see if this one is not more elegant and logicly fine (or not) 1516462910 N * Bertl Bertl_oO 1516477243 J * Aiken ~Aiken@2001:44b8:2168:1000:b26e:bfff:fe2a:b951 1516485501 Q * Riviera Server closed connection 1516485502 J * Riviera Riviera@2a03:b0c0:1:d0::10:b001 1516487399 M * arekm delta-ts64-fix01.diff builds on x86_64, now checking i686 1516492036 M * arekm Bertl: regular patch + delta-ts64-fix01.diff builds on x86_64 and on i686 (4.9.77) 1516492278 M * Bertl_oO great! thanks for testing!