IG-88

ja, hängt ein wenig von den karten ab, bei 1G fast alles, 10G recht viel, 2.5G intel geht garnicht und 2.5G realtek bei machen nur mit neuem treiber (den es nicht für 3615/17 gibt) sicherer du schreibst mal was für ein nic du planst bei DSM HA solltest du mal im forum suchen, da gab es zumindest leute mit problemen, ob das grundlegend ohne eine gültige SN läuft weiß ich nicht wenn du HA nachen willst dan würde ich eher 3617 empfehlen, keine ahnung ob das mit 918+ überhaupt geht nein, wenn die keinen lsi sas controller benutzt bist du nicht betroffen, so wie du das jetzt hast/planst ist alles ahci und das ist sicher test ist es irgendwie immer, es gibt da keine stable version aber wenn sie nicht ganz neu ist und nichts über probleme steht ist alles ok probleme kanns in der regel nur mit einzelenen treibern geben geundlegen braucht man für jeden loader typ eine andere extra.lzma aber du kannst ja erst mal mit dem default vom loader versuchen

https://xpenology.com/forum/forum/91-additional-compiled-modules/ (loader 1.02b dsm 6.1 with that extra's version number)

as its hard to see what hardware was used its easier that way then just trying in general the 1.03b loader needs legacy bios or csm mode if its a uefi bios i case of uncertainty's try loader 1.02b (dsm 6.1) that one work legacy and uefi and if that checks out with drivers you can use the gained knowledge how to set up the laoder to try 1.03b again and rule out other problems with preparing the loader or drivers

what controller? the normal onboard sata can't be a problem, if its a lsi sas then 918+ can have issues when disk hibernation is active imho 3617 is a better choice as it has some newer drivers then 3615 and supports more cpu cores, just a better 3615 2-3 years ago there might have been some packages that where available for 3615 bot not showed for 3617, that should not be much of a thing now and if you have the spk file at hand you can change the INFO file make it work i'd say make sure the hardware is working by using 3615 or 3617, if you don't need hardware transcoding and nvme you are fine with 3617 (its even better suited for ssd raid's as of F1 mode that 918+ does not have)

virtio_input is not used by synologys virtual dsm and i've not stumbled upon it yet (i never used kvm) integration virtio into the loader is on my list, i added virtio_input beside this i even don't know what the 9p modules are good for but i guess its enough to find out in what succession they are needed to load to make a test version afair there was some information missing for modprobe to work but maybe i'm wrong at all or that changed in the last 2 years i guess you wanted to mount the usb or the *.img file in windows to access the extra.lzma there is something in tutorials https://xpenology.com/forum/topic/29872-tutorial-mount-boot-stick-partitions-in-windows-edit-grubcfg-add-extralzma/ the flaws in that can be fixed so that win10 can access it the normal way, i used MiniTool Partition Wizard Free, converted the loader to MBR and back to GPT, that fixes the problem (its not much tested, not elegant , so its not mentioned anywhere yet, i would prefer a way with some open source tool but it was not that important as i fixed it on my usb's i use for testing) you could use a usb or just mount the img, "mount" is the thing you are looking for maybe just use a ready made usb and moount that to access extra.lzma i will have some time this weekend to work on extra's and drivers, maybe i should do a in between version with virtio and some newer realtek drivers

the new realtek drivers did not compile aginst old kernel 3.10.105 of 3615/17 the only new i could make would be for 918+ that would be the high end 10th gen, tested only once unsuccessful would you be willing to test a new patched driver to see if we can get the /dev/dri working (loader 1.04b 918+?) yours is 0x9BC5, UHD Graphics (Comet Lake 3x8 GT2) we only used one device for patching 0x3E92, UHD Graphics 630 (Coffeelake 3x8 GT2) we could try different one's as base that are present in the driver like 0x3EA5, HD Graphics (Coffeelake 3x8 GT3) 0x3EA0, UHD Graphics (Whiskey Lake 3x8 GT2) 0x3EA2, UHD Graphics (Whiskey Lake 3x8 GT3) maybe the Whiskey Lake UHD gets it up and running do you have a spare disk you could use for test install 918+? (you could test the newer 8122 driver too that way)

that say's ivy bridge and that would be one below 4th gen haswell (4th gen min. for 918+) https://ark.intel.com/content/www/us/en/ark/products/65735/intel-xeon-processor-e3-1220l-v2-3m-cache-2-30-ghz.html

i guess THATS the problem - 6.2.1 and 6.2.2 are incompatible to the drivers on the loader (6.2.0) the change was reverted in 6.2.3 use 6.2.(0) or 6.2.3 to install with the loaders default extra/extra2 (additional drivers are in these files)

bei "Marvell 88SE9215+JMicron JMB5xx" ist jedem gleich klar auf was das hindeutet mit "PCE8SAT-M01" findet man auch sachen die genauer sind https://mattgadient.com/8-port-sata-on-a-pcie-1x-lane-looking-at-the-pce8sat-m01-expansion-card/ JMB575 https://www.jmicron.com/products/list/16 1 zu 5 Port multiplexer, mach aus einem der ports des marvell 5, also 3 + 5 = 8 für dsm ist das nur eine 4port karte (wenn du von den 5 multiplex ports den ersten rausfindest (einfach probieren) sieht ok aus, bei jmb585 ist es eindeutig, das ist eine chip pcie-> sata mit 5 ports https://www.jmicron.com/products/list/15 das sollte mit den jmb585 karten kein problem sein (bei den asm1166 sindet dsm 32 pots pro karte und das gibt lustiges nummern chaos, geht an sich kann aber etwas verwirrend sein) aber ich dokumentiere mir auch immer die seriennummer der platten und tausche aus wenn das das system aus ist, da kann man nicht vesehentlich die falsche austauschen, wenn man die falsch rausnimmt merkt man das an der SN und packt sie zurück ohne das es folgen hat (ist ja alles stromlos) ... [Fri Jan 15 11:38:18 2021] pci 0000:00:1f.2: [8086:8d02] type 00 class 0x010601 ... [Fri Jan 15 11:38:19 2021] ahci 0000:00:1f.2: AHCI 0001.0300 32 slots 6 ports 6 Gbps 0x3f impl SATA mode [Fri Jan 15 11:38:19 2021] ahci 0000:00:1f.2: flags: 64bit ncq pm led clo pio slum part ems apst [Fri Jan 15 11:38:19 2021] scsi host0: ahci [Fri Jan 15 11:38:19 2021] scsi host1: ahci [Fri Jan 15 11:38:19 2021] scsi host2: ahci [Fri Jan 15 11:38:19 2021] scsi host3: ahci [Fri Jan 15 11:38:19 2021] scsi host4: ahci [Fri Jan 15 11:38:19 2021] scsi host5: ahci ... erster controller (onboard) mit 6 ports, die pci dev id "8086:8d02" kann man nachschlagen https://devicehunt.com/view/type/pci/vendor/8086/device/8D02 ... [Fri Jan 15 11:38:18 2021] pci 0000:03:00.0: [1b4b:9215] type 00 class 0x010601 ... [Fri Jan 15 11:38:20 2021] ahci 0000:03:00.0: AHCI 0001.0000 32 slots 4 ports 6 Gbps 0xf impl SATA mode [Fri Jan 15 11:38:20 2021] ahci 0000:03:00.0: flags: 64bit ncq sntf led only pmp fbs pio slum part sxs [Fri Jan 15 11:38:20 2021] scsi host6: ahci [Fri Jan 15 11:38:20 2021] scsi host7: ahci [Fri Jan 15 11:38:20 2021] scsi host8: ahci [Fri Jan 15 11:38:20 2021] scsi host9: ahci ... 4port controller, 88SE9215 sonst gibts da nichts weiter, sieht alles normal aus aber das deckt sich nicht mit deinen aussagen und sata port mapping der onboard controller hat nur 6 (intel chipset) und sonst sieht man da nichts von 2 weiteren ports check mal das handbuch (link?), theoretisch können ältere controller (in dem fall 2port) im bios zwischen ide und ahci modus wechseln, im ide modus würde er nicht als sata/ahci angezeigt aktuell so wie dmesg das jetzt hat wäre sata port map 64 sollte vom board noch was dazu kommen müsste man noch prüfen wo der sich einreiht, kann 624 oder 642 werden und wenn du den 9215 durch jmb585 ersetzt wurd aus der 4 eine 5 der 9215 ist selbst als 4port controller nicht so nützlich da er mit einer pcie lane bei pcie 2.0 nur 500MB/s für alle 4 port hat, das wird selbst normale hdd's ausbremsen (aber mit 1-2 genutzte ports kann man ihn schon nehmen) ... [Fri Jan 15 11:38:37 2021] drm: Unknown symbol fence_free (err 0) [Fri Jan 15 11:38:37 2021] drm: Unknown symbol fence_init (err 0) [Fri Jan 15 11:38:37 2021] drm: Unknown symbol fence_release (err 0) [Fri Jan 15 11:38:37 2021] drm: Unknown symbol fence_remove_callback (err 0) [Fri Jan 15 11:38:37 2021] drm: Unknown symbol fence_signal_locked (err 0) [Fri Jan 15 11:38:37 2021] drm: Unknown symbol fence_add_callback (err 0) [Fri Jan 15 11:38:37 2021] drm: Unknown symbol fence_enable_sw_signaling (err 0) [Fri Jan 15 11:38:37 2021] drm: Unknown symbol fence_signal (err 0) [Fri Jan 15 11:38:37 2021] drm: Unknown symbol fence_default_wait (err 0) [Fri Jan 15 11:38:37 2021] drm: Unknown symbol fence_context_alloc (err 0) [Fri Jan 15 11:38:37 2021] drm_kms_helper: Unknown symbol drm_crtc_add_crc_entry (err 0) [Fri Jan 15 11:38:37 2021] drm_kms_helper: Unknown symbol drm_framebuffer_cleanup (err 0) ... 918+ 6.2.3 mit jun's orginal extra/extra2, das vermurkst den i915 treiber für hardware transcoding (/dev/dri) wenn du das brauchst nimm die extra/extra2 von hier (die enfernt den treiber und sorgt dafür das der neue treibe läuft den dsm mitbringt) https://xpenology.com/forum/topic/28321-driver-extension-jun-103b104b-for-dsm623-for-918-3615xs-3617xs/die

and you need to reboot with that enabled to see the non uefi (legacy) usb boot device if csm is active and you choosed the uefi usb boot device then the csm will have no effect i think thats not important in our case

you just need to boot with loader 1.03b for 3615 and it will offer to migrate to the new system, it will use all settings and (if possible) plugins, also your date in volumes will remain the way they are no, for one 1.03b only hat one file (extra.lzma) and the default extra.lzma (meant to be used for 6.2.0) is also working for 6.2.3 and should have working drivers for the intel based nic you would only use the newer 6.2.3 extra.lzma (for 3615) when something does not work the new 6.2.3 extra/extra2 for 918+ is mandatory because of jun's added i915 drivers, 3615/17 don't have these drivers

i cant remember the code in the loader, maybe it only copies whats actually in rc.modules you can copy the files to into update but they wont be loaded as long as you dont do any changes to DSM kind of normal, there is nothing that would load "unknown" modules did you read the error message? modprobe is not supported/present in DSM DSM is a appliance and misses some things that might be seen as basic or normal insmod can be used for loading modules but you would need to care for dependencies yourself and load modules in the right order (usually modprobe takes care of this) the modules for virtio in the extra.lzma where meant to be used for testing by experienced people its also possible to unpack the extra.lzma, change rc.modules to load them and repack the loader, and use the repacked extra.lzma with the loader any (live) linux can do that with just lzma and cpio, you don't need any fancy chroot or kernel source, its just simple repacking the files as in the tutorial i wrote look in the bottom part 2. modify the "synoboot.img" https://xpenology.com/forum/topic/7187-how-to-build-and-inject-missing-drivers-in-jun-loader-102a/

try mbr loader version https://xpenology.com/forum/forum/98-member-tweaked-loaders/ maybe 1.02b is best for testing as its working with uefi and csm/legacy mode https://xpenology.com/forum/topic/13333-tutorialreference-6x-loaders-and-platforms/

0.12 was just a test version for something, as you found out already 0.11 has the files too you can extract the files manually and copy them to your dsm system for testing

maybe boot a rescue/live linux and do a "lspci -k" and post it here, that way we would see the used hardware in detail and the used drivers the cpu is to old for loader 1.04b / 918+ so start with 1.03b 3617 beside the link from above you should read the faq https://xpenology.com/forum/forum/83-faq-start-here/ and the tutorail made for 6.1 also applies to 6.2, with just a newer loader https://xpenology.com/forum/topic/7973-tutorial-installmigrate-dsm-52-to-61x-juns-loader/

open the extra.lzma with 7zip and see for yourself (extra3615_v0.12_test.zip), in /usr/lib/modules are the *.ko files in theory you should find the same files on you dsm system in /usr/lib/modules/update/ as the loader does a compare (every boot) between content of extra.lzma's *.ko files and whats in the update directory

over tumb rule, add all disks and remove biggest disk faq https://xpenology.com/forum/topic/9394-installation-faq/?do=findComment&comment=81094

maybe recycle bin is active?

voltage regulators can overheat too, also check for https://en.wikipedia.org/wiki/Capacitor_plague the psu can also cause that kind of problem

was für eine karte ist das genau? der 88SE9215 hat nur 4 ports und nutzt auch nur eine pcie 2.0 lane ( = max. 500MB/s) bei 8 ports ist entweder was mit multiplexer im spiel (da dann kann dsm nur 4 ports des eigentlichen chips finden/nutzen, die dahinter über sata multiplex kann dsm nicht) oder es gibt zwei von den chips und vor den beiden müsste eine pcie bridge sitzen die die beiden pcie lanes der controller mit dem board verbindet hatte ich mal hier was zu geschrieben (sind aber nur von begrenztem wert da man bei 2 pcie 2.0 lanes max 1000MB/s für 8 sata ports hat) https://xpenology.com/forum/topic/19854-sata-controllers-not-recognized/?tab=comments#comment-122709 ich rate zu pcie 3.0 fähigen chips mit 2 lanes wie dem jmb585 (5 port) oder asm1166 (6 port), wobei der asm bei mir eigenartige sachen macht, ich bentuze jmb585 habe ich hier ein wenig zusammengefasst https://xpenology.com/forum/topic/35882-new-sataahci-cards-with-more-then-4-ports-and-no-sata-multiplexer/ normalerweise musst du an dem sata portmap nichts verstellen kannst du ein /var/log/dmesg ziehen und posten, man könnte in dem log sehen was gefunden wird und wie sich die ports verteilen

the system partition (1st on every disk) is always a raid1 over all disks, nothing fancy SHR is just mdadm combined with lvm2, i guess its not going to work with auto detection but you can assemble the raids and then string them together with lvm2, you would need to have some knowledge about how SHR works and lvm btrfs should be in every recent linux distribution

yes, but all the usual limits a re there too all about raid modes and raid extension can be checked on synology#s help and faq/kb

you can open the extra.lzma with 7zip and check rc.modules and the *.ko files in "modules" HPE uses broadcom or intel and both are there you can also use iLO's virtual com port to check you the serial console using putty, that would give you a hint about whats going on i tested the hpsa.ko driver with a p410 lately, the p410 can only use raid mode so i created a raid0 and raid5 to test if its working and it did, so maybe start with that in theory the p420 can be switched to hba mode and in theory the driver should work with that too but as the p410 does not support that, i could not test this mode just don't assume things when you can test it -> see above, check fot the *.ko and in rc.modules - it is there (just the basic version form the kernel) so if something goes south you would need to check by serial console to pin it i cant do anything as i dont have a p420 to test with where did you get thsi or how did you create this hpsa.ko? if you want to know how to mod the extra.lzma you might need to read this https://xpenology.com/forum/topic/7187-how-to-build-and-inject-missing-drivers-in-jun-loader-102a/ just to check if its working or not you can copy the *.ko file todisk and load the driver manually with insmod but the driver is working (at least in raid mode) https://xpenology.com/forum/topic/36135-how-to-install-on-hp-dl380-g7/?do=findComment&comment=186543 use serial cable or iLO's virtual com port to check whats going on when booting the system, the only way to really know whats going on

imho you need to overlap internal and usb to have the usb disks as internal, like this maxdisks="24" esataportcfg="0x0" usbportcfg="0xfff000" internalportcfg="0xffffff"

have a backup and don't restart? i dont know your hardware but maybe a pcie ahci card and "peeling" the usb drives might be possible