Jump to content
XPEnology Community

nadiva

Member
  • Posts

    60
  • Joined

  • Last visited

Everything posted by nadiva

  1. not really, i didn't fill up the 2nd slot (1 NVME cost me 1/2 of NAS microserver) but i plan to upgrade to RAID0 as it fills up. if what you suspect is true (and RAID0 is also under risk), i'd consider adapter like Asus Hyper M.2 which i guess has a controller to face server with a bootable standalone RAID drive. Hopefully! If not, some shares will move back to spinning drives. Still the "intended purpose" for most people is the single member array setup i reckon, the speed and extra redundancy is worth it imo.
  2. this is a barebone setup with NVME setup as a standalone drive, same as SSD: Device Boot Start End Sectors Size Id Type /dev/nvme0n1p1 2048 4982527 4980480 2.4G fd Linux raid autodetect /dev/nvme0n1p2 4982528 9176831 4194304 2G fd Linux raid autodetect /dev/nvme0n1p3 9437184 4000795469 3991358286 1.9T fd Linux raid autodetect md4 : active raid1 nvme0n1p3[0] 1995678080 blocks super 1.2 [1/1] [U] ls /volume3 @eaDir Share1 @Share1@ homes @homes@ @quarantine @sharesnap @sharesnap_restoring @SnapshotReplication @synologydrive @tmp when placed on VM, speeds were horrible even with optimized drivers. This way, PCI3 is utilized to max.
  3. Yes they have nomenclatures, but only after patch (which is supposed to hide NVMEs via drivers). I think this will vary per model. My patched file wasn't overwritten by updates, but is still patched. Second limitation is in UI: Syno is not confident to use NVME arrays via their cheap hardware so they don't offer volume creation. We just do exactly what DSM would do during volume creation: use the same tools to create volume, and since then, it's treated equally.
  4. Since formatted with synopartition NVME acts exactly like other arrays, it has the the same small SYNO partition, and every UI/CLI disk related command works on it, from creating volumes, monitoring, trimming, replicating, share transfers back and forth. It had too much time to prove itself, and became the most reliable drive with highest availability along with SSD (even HDD RAID had to be rebuilt for no reason, just internal controller hiccup). Not bad for a cheap external PCI3x1-4. Once NVMEs are cheap, i will build big arrays from PCI5 NVMEs in order to utilize multi-40+gbit NICs In future, LAN if not WAN connection speeds must exceed local highend PC speeds. Not with official Syno boxes thou - their pathetic hardware is good for archaic 100Mbit samba only, i think they even use PCI2, that's why their UI pretends not to see NVME for volume creation, officially claiming lack of support "because of heat problems":)
  5. interesting topic, i never could make drives sleep, and it's frustrating from day one. never saw "ioctl device failed" errors, so my scemd is quiet. perhaps apparmor.log is the busy one, constantly dropping dmidecode messages dmidecode: AppArmor: /sys/firmware/dmi/tables/smbios_entry_point denied by profile /usr/sbin/dmidecode it's hard to test since hdparm -Y /dev/sdX will spin down volume, but it will wake up immediately: ata5.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6 ata5.00: waking up from sleep ata5: hard resetting link ata5: SATA link up 6.0 Gbps (SStatus 133 SControl 300) ata5.00: retrying FLUSH 0xea Emask 0x0 i wonder which nasty daemon is doing this? Dovecot stats? Marvell driver? Any of tons of DSM services? I don't know. In Proxmox, it'd be pvestatd, for example.
  6. my intended purpose too, but this was solved earlier in the thread. Running NVME as a standalone drive second year without issue (not even patching anymore with updates). The cache isn't cool idea for me, just like VM, i believe it gets reset after restart, allocates only around 400GB of multi TB drive. I prefer accessing all hot data from NVME and using RAID as a backup or replication target and cold data storage, and strongly prefer max speeds via 10Gbit NIC, where RAID seriously lags. Once you do the patch, you already win, then you just publish it into MD array to make it visible to DSM. Then do whatever you like in UI, e.g. move the shares to it.
  7. - Outcome of the update: SUCCESSFUL - DSM version prior update: DSM 6.2.3 25426-0 - Loader version and model: JUN'S LOADER v1.04b - DS918+ - Using custom extra.lzma: YES v0.11 - Installation type: BAREMETAL - HP ProLiant Microserver Gen10 X3421, 32GB ECC RAM, SDD. HDD, Intel NVME 2TB, Aquantia 10gbit NIC, DVD, USB Audio, UPS - Additional Comment: these updates are always easy, no patch for NICs or NVME main drive needed, just heard "system shutdown" and "system started" as badusb was typing TCG-Opal password in between.
  8. CALM DOWN coming back to the script as my NAS is busier and busier, and having now cpu load around 0.5, it can get noisy. The noise is sinuous and therefore annoying. There's no real load, it's just the amount of of services is high and fileindexd makes sure to make repeated load pops. i attach my fork for HP Gen10 (non Plus). core count is reported as 1 for AMD, instead for processors are there, using nproc have only 3 frequencies, so had to remove unused coolfreq didn't like IF for each core, double conditions, added it to the same block to save some cpu cycles instead added Ramp mode, to switch to Turbo mode faster if there's a load spike added debug mode "-d" deamon checks only once per 5 seconds, it doesn't make sense to check it too often, if the load stats isn't updated You can merge with the original and pick what you like. Thanks for the script as I now can have quiet machine again. 👌 note: High cpu equals Turbo, processing power (e.g. AES test, or 10gbit encrypted transfer) gets doubled with minimal freq raise, and is handled by cpu note: freq ramp and deramp will be slow just like the cat /proc/loadavg interval but this is fine for me, if a long operation (10gbit copy, backup, plex transcoding, emby maintenance) is going to take place, it will catch up soon and delay will be meaningless note: "Fan speed mode" in DSM might have zero impact, and by default my machine starts in power saving mode @ed_co : once scripts runs, it's in memory, as a basic concept, you need to kill the existing process and restart if you did modifications scaler.sh
  9. yeah so you install transcode in ikpgui an set it up in the script above: then insert another one...wait... another one...wait... just an example
  10. goal: have CD-ROM / DVD reader working, ready to server files asap without going to terminal, auto mount hardware: for example Microserver Gen10 with internal USB port and cute thin CD slot, it'd be a shame to leave it empty software: find cdrom.ko and sr_mod.ko from someone nice who can build initial setup here the work can end, if you like to operate it via CLI. Automount is crippled like on other appliances, we can hook up to udev, e.g. monitor actions here: normally udev would do the job, but here events are limited if not delayed, there's just change action, that's what i put a respective command to /etc/udev/rules.d/99-local.rules: load the rules: then process its events to make sure actions are captured, provide automount (sometimes delayed) and optionally make sound when CD is inserted or ejected via AudioStation (just grab mp3 fast on https://ttsmp3.com/), cdrom.sh: however it's even more crippled, the triggers are fired often too late (or only when un/mounting on CLI) and never when doing manual button oepration, so we best periodically launch blkid or eject -a on /dev/cdrom via scheduler. I choose to fire the latter to save some cpu cycles as it's faster and i launch it each 30 seconds. In the end, this way we avoid CLI completely, it gets un/mounted automatically with little delay (~15sec, loading CD alone is 10sec so i think it doesn't matter). > Some devices prevent manual buttons once mounted but users love them instead of CLI -> this is fixed during mount. > CD will be inserted with physical tray push > CD will be ejected with physical eject button OR if you love CLI: umount /dev/cdrom OR eject (no parameters)
  11. nadiva

    VPN

    openvpn -> wireguard for breathtaking speeds
  12. - Outcome of the update: SUCCESSFUL - DSM version prior update: DSM 6.2.2-24922-update 5 - Loader version and model: JUN'S LOADER v1.04b - DS918+ - Using custom extra.lzma: YES v0.11 - Installation type: BAREMETAL - HP ProLiant Microserver Gen10 X3421, 32GB ECC RAM, SDD. HDD, Intel NVME 2TB, Aquantia 10gbit NIC, DVD, USB Audio, UPS - Additional Comment: Semibrick install (PAT in UI + copy extras in file manager before boot). Lost few settings and root folder. Both NVME as standalone drive and 10gbit 3rd NIC worked after patches and another reboot
  13. Good to know. I am now rsyncing these folders now (what a cheap small backup). Maybe mirroring boot drive would be helpful too. I was scared to do "migration" instead of semibricking method because you say "In case of a "migration" the dsm installer will detect your former dsm installation and offer you to upgrade (migrate) the installation, usually you will loose plugins, but keep user/shares and network settings". I don't know what are "plugins" in DSM world but sounds scary. I hope you don't mean packages. This version 6.2.3 didn't solve any problem so far - CloudStation/Drive chaos, Moments/Photos chaos, replicating homes still crashing although explicitely "fixed" recently.. Can't wair for DSM7!
  14. I waited a bit and after 6.2.3 25426-0 PAT was available (today) I started the upgrade. HP Gen 10 with 10gbit NIC and standalone NVME drive, DVD, and USB sound. Upgrade from 6.2.2 extra918plus_v0.7_test -> 6.2.3 extra918plus_v0.11 was very easy: upgrade manually in UI (reboot) copy extras to USB drive in Windows and boot patch_LAN; activation patch (newest but not yet with this version, just add the ""6.2.3 25426-0"" string as last); libNVMEpatch (newest) patch and reboot Super quick, no rufus / osfmount / imager, lovely. NVME standalone drive - didn't remove it before upgrade despite recommendation (but i don't use it as a cache), and didn't need to set up "md" logical drive after upgrade. Just patch. Same with 10gbit Aquantia NIC. On boot, server speaks "System started", pushbullet "NAS" custom "device" sends notification but it deleted the custom notification as a way of making sure i'm aware of restart via all possible ways. It also deleted the /root folder so will spend a lot of time to reconstruct my hacks.
  15. actually 2$ speaker discharged every few weeks and sound was terrible. Unlike this 10$ USB sound card with integrated speaker, whichs sounds great! (and is the only of this kind i found, combining both sound card and USB powered on a single USB plug). Then you can add custom DSM UI notifications along with sound to your scripts (you want your own Notification events with email? sure: https://www.beatificabytes.be/send-custom-notifications-from-scripts-running-on-a-synology-new/😞 synodsmnotify @administrators "DSM notification" "Hello user" /volume1/@appstore/AudioStation/bin/mplayer /tmp/test.mp3 If you want MORE sound.. you add boot and shutdown script to make noise. Useful! Finally you want to add sound to standard Synology events. I do this. I translate the incoming emails in a dedicated folder to sound. Not with generic TTS technology (no such package found), but with custom sounds generated at http://www.fromtexttospeech.com/ . I care about all "failures" and few more events. So i just translated the text to mp3 quickly on that website. Then I run email harvester to play it based on specific email content: #/bin/bash id=$(id -u $1) echo "Monitoring user [$id] emails..." while true; do #collect list of emails arrived last minute find /volume1/MailPlus/@local/$id/$id/Maildir/.INBOX.Notifications/cur/ -type f -mmin 1 | while IFS= read -r newemail; do #grep error text and Synology header if [ "$(grep -Pzl '(?s)failed.*\n.*Synology DiskStation' $newemail)" == "$newemail" ]; then nohup /volume1/@appstore/AudioStation/bin/mplayer /root/failuredetected.mp3 2> /dev/null & echo "$(date) Generic failure alert" elif [ "$(grep -Pzl '(?s)An I/O error occurred to drive.*\n.*Synology DiskStation' $newemail)" == "$newemail" ]; then nohup /volume1/@appstore/AudioStation/bin/mplayer /root/driveerrordetected.mp3 2> /dev/null & echo "$(date) Drive I/O error alert" elif [ "$(grep -Pzl '(?s)A file system error was detected.*\n.*Synology DiskStation' $newemail)" == "$newemail" ]; then nohup /volume1/@appstore/AudioStation/bin/mplayer /root/filesystemerrordetected.mp3 2> /dev/null & echo "$(date) File system error alert" elif [ "$(grep -Pzl '(?s)was not ejected properly.*\n.*Synology DiskStation' $newemail)" == "$newemail" ]; then nohup /volume1/@appstore/AudioStation/bin/mplayer /root/usbnotejectedproperly.mp3 2> /dev/null & echo "$(date) USB ejection alert" fi done sleep 60 done You could also use one of the notification packages to forward notifications to specific email / text file and process it with different script. That'd add couple of more events (e.g. USB copy finished, why is this isn't emailed:(( ). This way we have so far: * AudioStation very nice playback * DSM boot, shutdown, custom tasks and any Synology events interpreted as speech 🗣️ or sound 🔊 * no external packages needed, no rechargeable batteriers needed, no QNAP NAS needed (feat. voice notifications, why Synology doesn't have it?), no more crying about missing "beep" DSM function on a custom hardware To do: * remote control * redirect Video Station to USB sound, add big screen and make it a 4K home cinema
  16. no need for gpu, those mentioned gpus are rather weak, it doesn't make much sense of processing just slightly faster. i only like the benefit of displayport vs VGA on Gen10. buying next server, i will not consider gpu at all, and unless i want to have 100W server, which i don't, i won't care about a weak gpu inside, which can't transcode 4K anyway.. just let the cpu process those files, it works well. not worth of moving to omv whatsoever. running 6.2.2 with 10gbit and standalone NVME, i still prefer it vs openmediavault/freenas big time. had 2 crashes (1 brtfs panic and 1 btrfs kworkerd freeze), not ideal, but no consequences like in ZFS which totally failed. having NVME sync to RAID creates tripple redundancy, while SSD holds apps. moved from replication to hyperbackup as homes couldn't replicate and browsing backups is comfortable in hyperbackup. found this trick if you hate syno shares (photo, video, music) and want to use your own folder structure mount --bind /volumeX/Group/Pictures/ /volumeX/photo/ unlike ln -s command, this one is not ignored by syno and is fully visible and processed by indexing. Add to rc.local. automatic (Lets Encrypt) certificate distributor: git new plex to sync (my fork actually works much more stable): git can't wait for DSM7 discussions, nothing really new coming but they fix Moments/PhotoStation trouble (was inconsistent by all means with other apps and not reliable).Still no FDE:(
  17. thanks for the guide. there are conditions for this communication, two way authentication should be temporarily off (error 402), host/port must be the same in the process, and in my case, couldn't use hostnames. Use private session not to store credentials to history.
  18. - Outcome of the update: SUCCESSFUL - DSM version prior update: DSM 6.2.2-24922-update 4 - Loader version and model: JUN'S LOADER v1.04b - DS918+ - Using custom extra.lzma: YES. IG-88's v0.7beta - Installation type: BAREMETAL - HP ProLiant Microserver Gen10 X3421, 32GB ECC RAM, Intel NVME 2TB + Aquantia 10gbit - Additional Comment: Manual PAT v5 install and reboot required. Both NVME as standalone drive and 10gbit 3rd NIC worked after boot without any fixes.
  19. beauty of ZFS/BTRFS beside copy-on-write, checksumming, is what most people don't use even when given for free: - snapshots which is comfortably visible as Previous Versions in windows Explorer properties - files not overwritten when replaced, deleted, instead kept as transaction or put to samba thrashbin (+snapshot +replication +Drive versioning +online backup) forums: - efficient operation with blocks, deduplication, instead of rsyncing files on high level, you can copy whole shares as it was a file, in ZFS without reencrypting - you can create a block device inside of a pool, the VM will think it's Ext4, while you add benefits of your fs on top of this primitive fs - compression with negligible impact which works on data partitions, i had 30%+ savings - (ZFSonLinux) native encryption, which i've tested as super fast on debian list of good resources: xpenology list of syno ports: https://www.synology.com/en-global/knowledgebase/DSM/tutorial/Network/What_network_ports_are_used_by_Synology_services list of syno commands: https://github.com/wuseman/SYNOLOGY my notes full of commands to push limits and break rules of DSM, freenas, OMV benchmarks notes (pic):
  20. No, i ment that those are the busiest times to produce noise. I don't think my GPU was ever used and honestly i don't care: Yes I actually never tried the cache and just formatted it into a volume due to - cache limit, not utilizing most of the space - i love main data to reside on high IOPS high sequential speed NVME drive, then replicate to RAID, that's why i bought it, not for a cache which canshould be substituted with RAM - i didn't move all stuff yet but later i want incremental backup with snapshots, ZFS/BTRFS style send->receive, and this is done with 'Snapshot replication' foremost or 'HyperBackup'. All other backups are amateurish, they don't send changed blocks (in ZFS even encrypted), and snapshots for double recovery from each drive, and even system files. This is where the beauty of ZFS/BTRFS is
  21. Gen10 finding. See my story here. Indeed after boot, performance is lower and fans never scream. I had to tweak it that way (now added as post boot script) in order to get the following improvement: openssl speed -elapsed -evp aes-256-xts from 1200000 to 3200000 OPENSSL_ia32cap="~0x200000200000000" openssl speed -elapsed -evp aes-256-gcm from 60000 to 160000 hdparm -Tt /dev/nvme0n1 from 1200 to 2500MBs dd if=/volume3/moviefile of=/dev/zero bs=4G count=1 from 600 to 1200MB/s same idle power consumption, but 3x faster calcs, 2x faster drives and some fan noise under load. If anyone has openssl xts last number 1mil instead of 3, you better check this. seems acpi-cpufreq driver, the only one available as seein in ls /usr/lib/modules/*cpu*, is confused here, and sets low TPU max (12W), but the CPU can boost to 35W when needed. Other cpu freq drivers (not governers) for AMD are powernow-k8 and pcc-cpufreq, the first one supposed to by replaced by acpi-cpufreq, the later one is "recomended for ProLiant servers". So i looked into my collection of DSM test files, but there isn't one. this also gives us idea how full disk encryption could be fast - up to 3.2GBs! Not far from i7 Pro desktop performance. I personally have disk encryption as fast as openssl benchmark (xts or core count * cbc/gcm bench) on any device incl router. (My router provided me encrypted 1gbit shares for years at full speed, that's why i find it a complete joke to buy a NAS which can do 1gbit encyption only.) here you have evidence how poor are all file encryptions. for those who dont believe how poor is file encryption, see the numbers: dd if=/dev/zero of=/volume3/normalsharedfolder/test bs=4G count=1 0+1 records in 0+1 records out 2147479552 bytes (2.1 GB) copied, 1.90531 s, 1.1 GB/s dd if=/dev/zero of=/volume3/encryptedsharedfolder/test bs=4G count=1 0+1 records in 0+1 records out 2147479552 bytes (2.1 GB) copied, 1.88765 s, 184 MB/s cached file copy same. samba surprisingly faster - 300MB/s. puzzling as no physical writes happen when copy done, no cache effect. still encryptfs is not good. Full disk encryption would do 1.0-1.1GB/s. No doubt Qnap doesn't use this. In the era of native encryption coming to ZFS, it's shocking Synology would use this obsolete method.
  22. is 1 ok for AMD? 4 cpus are listed in cpuinfo, but cores=1. interesting by default cpu is at max freq all of the time in all OS, but in DSM the performance is halved after boot (proven by benchmarks). scaling_setspeed is <unsupported>. Governor is performance. Min freq, max freq is correct. cat $(ls /sys/devices/system/cpu/cpufreq/policy*/scaling_cur_freq) 2100000 2100000 2100000 2100000 grep "MHz" /proc/cpuinfo cpu MHz : 2100.000 cpu MHz : 2100.000 cpu MHz : 2100.000 cpu MHz : 2100.000 cpuinfo_min_freq, cpuinfo_max_freq, cpuinfo_cur_freq, all seems correct, but performance is lower aes-256-xts 90198.74k 342936.66k 777926.06k 1132937.56k 1292978.86k setting governors does nothing, powersave keeps lowest freq, performance keeps highest freq (only the scaler.sh does the job, although very slowly.). driver is acpi-cpufreq, only one available on DSM. i set scaling_driver to highest freq, still low performance and fans off. dmidecode and UI shows highest freq. Performance still low. tried hundred commands, only when setting scalers up and down (which sets userspace), and then setting performance: for i in 0 1 2 3; do cpufreq-set -f 1400 -c $i; done for i in 0 1 2 3; do cpufreq-set -f 2100 -c $i; done for i in 0 1 2 3; do cpufreq-set -u 1400000 -c $i; done for i in 0 1 2 3; do cpufreq-set -u 2100000 -c $i; done for i in 0 1 2 3; do cpufreq-set -g performance -c $i; done ........performance is unlocked and fans go loud under load aes-256-xts 225339.77k 833591.27k 1901271.72k 2758488.41k 3190287.02k idle consumption is the same. i think the sequence of commands unlocked CPU TPU max from 12W to 35W. Crazy.
  23. so you can just use synodsmnotify @users title message, or enhance use with community packages "Growler", "Notification Messenger", "Notification forwared", "Notification Essentials". And add own sound to it. Ideally, i'd like a package which can read the message loud:) Anyways, here's the notification with sound (pic):
  24. overlooked this post, indeed i did this meanwhile and running NVME drive happily. Now i'm searching for a solution to turn on the full disk encryption and idle the spinning drives. like it was possible years ago, when the procedure was easy: install DSM without spinning drives, add them later. Nowadays drives are resumed immediately after hdparm -Y <drive>.
  25. yep, thanks for that. heh i might buy a syno for family, but for me i wouldn't call their retail machines great, as they throw outlet CPUs inside, no ECC, no full disk encryption, security is poor out of the box, running services under root, forcing Plex to be an user sharing the data with real users (who are forced to be part of users group), etc etc. but but all of this can be tweaked in a persistent way. same with the patches - i find it trivial to make it persistent even after upgrades. even other OS deleted some of my tweaks, actually exactly the same - e.g 10gbit NIC, so i scheduled it using their persistent scheduler. regarding NVME, hearing people have no problem running after minor and major upgrades! that's why i love DSM on custom machine. replace rubbish hardware with decent one, harden the security, and enjoy the greatness. re noise, i'm lucky with it, but it seems DSM is throttling the speed down as well. all seems to be fast generally, but e.g. openssl speed is not fast as it should be. no impact of fan speed mode in Hardware & Power panel. also no luck with synofandebug, synofanconfig, synotune. oh wait, why do i have memory compression on! re audio, yes it works, plug in some 5$ usb sound card with $2 speakers attached, and AudioStation will play music from your custom machine asap. to test, use: /volume1/@appstore/AudioStation/bin/mplayer <file.mp3> and now that we have this command, we can load a notification package from one of the two know alt repositories, and ta da... you have custom notifications with sound. 🔊 i'm customizing what i can: plex trakt total sync for free, ssh via certificate even with homedrives unlocked, mandatory parity check postpone, shell with autocomplete/autocorrect/syntaxhighlightning, update certificate automatically into all apps and make it valid for external domain and on LAN at the same time without ever confirming in browser! etc. if you're interested, let me know.
×
×
  • Create New...