If a failure occurs ...


Recommended Posts

Bonjour,

 

Mon serveur (comme inscrit dans ma signature) est pret bien que je n'ai pas encore acheté de licence.

 

Je trouve ce système assez bon (bien que le système de plugins soit pas top, trop fouillis, pas pratique. Les plugins que j'ai tenté d'installer ne s'installait pas ou ne fonctionnait pas)

 

Mais mon interrogation est ailleurs. La phylosophie d'unRAID qui ne donne pas le choix du type de configuration des disques est un peu déroutant ; On aurait pu avoir le choix, pour le prix ça aurait été la moindre des choses.

 

Mais puisque le choix est fait, j'aimerais comprendre.

 

Car si le site dit qu'en cas de panne, ON NE PERD PAS TOUT, j'ai un autre serveur en RAID1 et en cas de panne JE NE PERDS RIEN. Que se passe-t-il si la parité est trop petite et qu'un gros disque tombe en panne ? On perd énormement de données. Pas en RAID ....

 

Sur une version licenciée Pro, on a pas de limite de disques, c'est très bien mais peut-on avoir le nombre de parité que l'on souhaite ? Histoire de limiter les problème en cas de panne.

 

Merci de m'eclairer, cela déterminera certainement mon achat.

__________________________________________________________________________________________________________________________________________________________

 

Hello,

 

My server (as written in my signature) is ready although I have not yet bought a license.

 

I find this system pretty good (although the plugin system is not top, too messy, not practical.The plugins I tried to install did not install or did not work)

 

But my question is elsewhere. The unRAID phylosophy which does not give the choice of the type of configuration of the disks is a bit confusing; We could have had the choice, for the price it would have been the least of the things.

 

But since the choice is made, I would like to understand.

 

Because if the site says that in case of a breakdown, we do not lose everything, I have another server in RAID1 and in case of breakdown I DO NOT LOST. What happens if the parity is too small and a large disk fails? You lose a lot of data. Not in RAID ....

 

On a licensed version Pro, we have no disk limit, that's fine but can we have the number of parity we want? History to limit the problem in case of breakdown.

 

Thank you for enlightening, this will definitely determine my purchase.

Link to comment

The key concept with unRaid vs other RAID systems (RAID 1, RAID5, RAID10, RAID50) is that if you exceed the tolerance of failures on a true RAID system you will have lost 100% of all of your dataGuaranteed

 

On unRaid, if you exceed its tolerance level (ie: lose a total of 3 disks), you will only have lost the files which were stored on the disks which died.  All other disks still are functional with all of the files contained on them intact.

 

It all depends upon what you're comfortable with.  If you lose more disks than the tolerance level do you want to lose 100% of the files or only a portion?

 

The parity disk(s) are always going to be as large or larger than the largest data disk.  No choice about that.  And you have the option to use either 1 or 2 parity disks which sets the tolerance level of the number of disks which can concurrently drop dead without suffering any data loss (either 1 or 2 disks)

 

https://lime-technology.com/wiki/index.php/Parity

 

As far as the plugin system is concerned, what were you trying to install?  Generally, the only plugin you need to install via the plugin system is Community Applications.  After that, you install everything from the Apps tab which will only ever display items which are compatible with your particular version of unRaid

Link to comment

Si la parité ne peut être que le plus grand disque disponible, que se passe-t-il si la configuration

 

parité (1To) disk1 (500 Go) disk 2 (250 Go)

 

devient

 

parité (1To) disk1 (500 Go) disk 2 (250 Go) disk3 (2 To)

 

La système va m'imposer mon nouveau disque en parité et non en stockage ? On ne ferais donc pas ce que l'on veut de ses disques si c'est le cas.

 

Si on peut l'attribuer en disk3, la parité ne suffit plus pour assurer la sécurité totale des données ; Dans les 2 cas, c'est pas top.

 

Sinon, jee reformule ma question (Je ne trouve pas l'info) :

 

Si j'ai 10 disques durs (tous de la même taille ou étant par paires identiques), puis-je en attribuer la moitié en parité ?

 

Et personnellement en 20 ans d'utilisation du RAID, je n'ai jamais perdu de données (il faut simplement éviter des disques durs venant d'un même lots).

__________________________________________________________________________________________________________________________________________________________

 

If the parity can only be the largest available disk, what happens if the configuration

 

Parity (1TB) disk1 (500GB) disk 2 (250GB)

 

Becomes

 

Parity (1 TB) disk1 (500 GB) disk 2 (250 GB) disk3 (2 TB)

 

The system will impose my new disk in parity and not in storage? We would not do what we want with his records if that's the case.

 

If it can be assigned to disk3, parity is no longer sufficient to ensure total data security; In both cases, it is not top.

 

Otherwise, I rephrase my question (I can not find the info):

 

If I have 10 hard drives (all of the same size or being in identical pairs), can I assign half as a parity?

 

And personally in 20 years of use of the RAID, I never lost data (it is necessary to avoid hard disks coming from the same lots).

Link to comment

 

If the parity can only be the largest available disk, what happens if the configuration

 

Parity (1TB) disk1 (500GB) disk 2 (250GB)

 

Becomes

 

Parity (1 TB) disk1 (500 GB) disk 2 (250 GB) disk3 (2 TB)

 

The system will impose my new disk in parity and not in storage? We would not do what we want with his records if that's the case.

 

If it can be assigned to disk3, parity is no longer sufficient to ensure total data security; In both cases, it is not top.

 

Otherwise, I rephrase my question (I can not find the info):

 

If I have 10 hard drives (all of the same size or being in identical pairs), can I assign half as a parity?

 

And personally in 20 years of use of the RAID, I never lost data (it is necessary to avoid hard disks coming from the same lots).

Parity always has to be as large or larger than the largest data disk.  Any deviation from that and the array will not start since it's an invalid configuration.

 

You can however have disks outside of the parity protected array that can be any size you choose (but obviously they won't be protected from a failure)

 

Nor have I ever suffered any data loss.  But I'm of the opinion that if its remotely possible to lose data, I'd rather lose some but not all.  unRaid pretty much guarantees I'd only ever lose some, but never all in the absolute worst case scenario.  RAID guarantees that in the worst case scenario I'd lose everything.

 

Another thing to consider in these days of monster sized drives is the rate of URE's on the drives. 

 

A URE on a true raid system during a rebuild constitutes a drive failure.  A URE on unRaid does not constitute a failure.  Worst case scenario with a URE on unRaid when you already have 2 dead drives is that a single file might get corrupted.  Same situation on a true RAID system results in the loss of the complete array.

 

Link to comment

J'ai regardé mais j'avoue ne pas avoir tout compris (mon anglais est très mauvais)

 

La chose que je voulais savoir c'est que sur 10 disques, je pouvais en avoir 5 en parité, si la réponse est non uRAID est faillible pour moi.

 

Je comptais monter à terme un serveur de 10x6 To dont 5 en parité. Si cela n'est pas possible et que seuls 2 disques peuvent être en parité. La tolérance de panne me semble trop élevée et bien moins sûr qu'un RAID1

 

__________________________________________________________________________________________________________________________________________________________

 

I looked but I confess I did not understand everything (my English is very bad)

 

The thing I wanted to know is that out of 10 drives, I could have 5 in parity, if the answer is no uRAID is fallible to me.

 

I expected to eventually upgrade a server of 10x6 TB with 5 in parity. If this is not possible and only 2 discs can be in parity. The fault tolerance seems too high and much less safe than a RAID1

Link to comment

Ceci ne répond pas à la question :)

 

Ou je dois comprendre que aRAID est incapable de gérer plus de 2 parités et serait donc limité.

 

Combien de parités unRAID sait-il gérer ?

 

__________________________________________________________________________________________________________________________________________________________

 

This does not answer the question :)

 

Or I have to understand that aRAID is unable to handle more than 2 parities and would therefore be limited.

 

How many parities does unRAID handle?

Link to comment

Ceci ne répond pas à la question :)

 

Ou je dois comprendre que aRAID est incapable de gérer plus de 2 parités et serait donc limité.

 

Combien de parités unRAID sait-il gérer ?

 

__________________________________________________________________________________________________________________________________________________________

 

This does not answer the question :)

 

Or I have to understand that aRAID is unable to handle more than 2 parities and would therefore be limited.

 

How many parities does unRAID handle?

Up to 2 parity disks.  Which could then handle 2 concurrent drive failures within the array before any loss of data.

 

Your example stated 10 disks including 5 parity disks which could handle 5 concurrent failures.

 

Unless you're misunderstanding parity.  On a true RAID system, the parity data is distributed among all of the array disks (and the data is striped across all the disks also).  unRaid uses dedicated parity disks (similar to RAID4), but does not have the data striped across the array disks (which is what lets unRaid carry on if more than two disks die concurrently unlike a true RAID system)

Link to comment

2 parity drives as already stated.

 

But 2 parity drives will let you recover from 2 simultaneous failures. And if you have more than 2 simultaneous failures then only the particular data disks that failed are affected. Each data disk is a completely independent filesystem. And even if you have more than that many failures simultaneously the data from the disks can often be recovered with filesystem recovery tools as long as the disk can be read.

 

The independence of the data disks while still providing parity protection is really what unRAID is all about.

Link to comment

OK et 2 disques (de 6 To) peuvent être suffisants pour 8 autres disques de 6 To ?

 

J'ai quand même des doutes (surtout si le taux de remplissage des disques est élevé)

 

____________________________________________________________________________________________________________________________________________________________

 

OK and 2 discs (6TB) may be enough for 8 other 6TB drives?

 

I still have doubts (especially if the filling rate of the discs is high)

Link to comment

I've re-read the posts, and I'm going to change my answers around.  (Something might be getting lost on the translation)

 

unRaid can have something like 28 data drives possible.  Using 2 parity disks, unRaid can recover from 2 simultaneous drive failures on those data drives before any data loss happens.

 

If more than two disks out of the 28 drives die at the same time, then you will have lost a portion of the files, but not all of them.

Link to comment

OK et 2 disques (de 6 To) peuvent être suffisants pour 8 autres disques de 6 To ?

 

J'ai quand même des doutes (surtout si le taux de remplissage des disques est élevé)

 

____________________________________________________________________________________________________________________________________________________________

 

OK and 2 discs (6TB) may be enough for 8 other 6TB drives?

 

I still have doubts (especially if the filling rate of the discs is high)

Please read the link I gave.

 

Parity by itself does not allow data recovery. This is true even of RAID systems. Parity just provides bits that allow the data from a missing data disk to be calculated from all the other disks. So the number of data disks doesn't matter, just the number of simultaneous failures. If you only have one failure then one parity plus all the other disks can be used to calculate the missing disk's data.

 

The filling rate is also irrelevant since any system, whether RAID or unRAID, will only let you write data as fast as it can be written. Since unRAID does not stripe data. and each file is completely on a single disk, then reads will only happen at single disk speed, and writes will be somewhat slower than that since parity must also be calculated and written.

 

If you need more speed then btrfs cache pools are available. All of this is covered at that page I linked you to. If you read that then maybe you can ask better questions.

Link to comment

Je comprends mieux. On peut comparer ce système à celui de PARchive (Linux) et QuickPAR (Windows), sauf que ces derniers travaillent au niveaux des fichiers et pas des disques eux-mêmes.

 

Ça devient plus clair.

 

Une question me vient de ce fait :

 

Si ma grappe évolue en taille, je peux sortir un disque de parité et le remplacer par un de plus grande taille ?

 

__________________________________________________________________________________________________________________________________________________________

 

I understand better. One can compare this system to that of PARchive (Linux) and QuickPAR (Windows), except that they work at the file level and not the disks themselves.

 

It becomes clearer.

 

One question comes to me from this fact:

 

If my cluster grows in size, can I get a parity disk and replace it with a larger one?

Link to comment

Je comprends mieux. On peut comparer ce système à celui de PARchive (Linux) et QuickPAR (Windows), sauf que ces derniers travaillent au niveaux des fichiers et pas des disques eux-mêmes.

 

Ça devient plus clair.

 

Une question me vient de ce fait :

 

Si ma grappe évolue en taille, je peux sortir un disque de parité et le remplacer par un de plus grande taille ?

 

__________________________________________________________________________________________________________________________________________________________

 

I understand better. One can compare this system to that of PARchive (Linux) and QuickPAR (Windows), except that they work at the file level and not the disks themselves.

 

It becomes clearer.

 

One question comes to me from this fact:

 

If my cluster grows in size, can I get a parity disk and replace it with a larger one?

I just want to clarify something before you get the wrong idea.  PAR is used to recover corruption within a file.  unRaid's (or any raid system for that matter) does not recover corruption on a disk.  Rather they are used to recover (rebuild) a drive that has failed in operation.

 

The filesystem tools itself (XFS / btrfs) are used to recover from any file system corruption.  (Most likely cause is a loss of power during a write to the array - get a UPS to solve this problem)

 

Link to comment

I don't want to discourage you from a good product, but from the conversation so far, I have doubts about your success with unraid. You may wish to email limetech directly [email protected] and ask about support options in your language. The forum is english only as you found out, and lack of english fluency will limit your ability to get good help here on the forums without getting frustrated.

 

You will have to be patient with us on the forums as we try to help you, and likewise you will likely have to wait longer for a response as it can be frustrating communicating with someone who has trouble understanding what we are trying to tell you.

 

Please understand I want you to have a good experience with unraid, but language may be an issue.

Link to comment

If my cluster grows in size, can I get a parity disk and replace it with a larger one?

Yes you can replace parity with a larger one. And you can replace a data disk with a larger one also. So one way you can increase capacity is to add disks, and another way to increase capacity is to replace existing disks with larger disks.

 

unRAID will rebuild the contents of the original smaller disk onto the new larger replacement disk. The only restriction is that parity must be at least as large as the largest single disk, so there will be parity bits to cover the size of the largest disk.

 

And you can do this one disk at a time because it doesn't matter if the data disks aren't the same size since each data disk is independent.

 

Link to comment

I don't want to discourage you from a good product, but from the conversation so far, I have doubts about your success with unraid. You may wish to email limetech directly [email protected] and ask about support options in your language. The forum is english only as you found out, and lack of english fluency will limit your ability to get good help here on the forums without getting frustrated.

 

You will have to be patient with us on the forums as we try to help you, and likewise you will likely have to wait longer for a response as it can be frustrating communicating with someone who has trouble understanding what we are trying to tell you.

 

Please understand I want you to have a good experience with unraid, but language may be an issue.

 

Aucun soucis, j'arrive à comprendre (et j'espère me faire comprendre aussi lol).

 

Tout va bien.

 

La phylosophie de fonctionnement m'echappais un peu mais je commence à en comprendre le fonctionnement. Bon, je n'en ferais pas un serveur principal de sauvegarde, j'aurais un autre NAS (d'une autre marque) en RAID6 comme mirroir réseau avec RSync.

__________________________________________________________________________________________________________________________________________________________

 

No worries, I manage to understand (and I hope to make me understand also lol).

 

Everything is fine.

 

The operating phylosophy escaped me a little but I begin to understand how it works. Well, I would not make it a primary backup server, I would have another NAS (from another brand) in RAID6 as a network mirror with RSync.

 

 

Je comprends mieux. On peut comparer ce système à celui de PARchive (Linux) et QuickPAR (Windows), sauf que ces derniers travaillent au niveaux des fichiers et pas des disques eux-mêmes.

 

Ça devient plus clair.

 

Une question me vient de ce fait :

 

Si ma grappe évolue en taille, je peux sortir un disque de parité et le remplacer par un de plus grande taille ?

 

__________________________________________________________________________________________________________________________________________________________

 

I understand better. One can compare this system to that of PARchive (Linux) and QuickPAR (Windows), except that they work at the file level and not the disks themselves.

 

It becomes clearer.

 

One question comes to me from this fact:

 

If my cluster grows in size, can I get a parity disk and replace it with a larger one?

I just want to clarify something before you get the wrong idea.  PAR is used to recover corruption within a file.  unRaid's (or any raid system for that matter) does not recover corruption on a disk.  Rather they are used to recover (rebuild) a drive that has failed in operation.

 

The filesystem tools itself (XFS / btrfs) are used to recover from any file system corruption.  (Most likely cause is a loss of power during a write to the array - get a UPS to solve this problem)

 

 

Oui je comprends bien la distinction.

 

Le format PAR(2) permet aussi de reconstruire des données absentes, pas forcement corrompues. Et dans le cas d'unRAID on est en présence de données absentes (puisque absence d'un disque dur) qu'il faut reconsctruire. Le principe est pas forcement différent.

 

Si l'utilisation de la partité est différente, la finalité est un peu identique: La récupération de données (qu'elles soient absentes ou corrompues) est le but premier ^^

__________________________________________________________________________________________________________________________________________________________

 

Yes, I understand the distinction.

 

The format PAR (2) also makes it possible to reconstruct data that is absent, not necessarily corrupted. And in the case of unRAID one is present of data absent (since absence of a hard disk) that must be reconstructed. The principle is not necessarily different.

 

If the use of the partite is different, the purpose is somewhat identical: The recovery of data (whether absent or corrupt) is the primary goal ^^

 

If my cluster grows in size, can I get a parity disk and replace it with a larger one?

Yes you can replace parity with a larger one. And you can replace a data disk with a larger one also. So one way you can increase capacity is to add disks, and another way to increase capacity is to replace existing disks with larger disks.

 

unRAID will rebuild the contents of the original smaller disk onto the new larger replacement disk. The only restriction is that parity must be at least as large as the largest single disk, so there will be parity bits to cover the size of the largest disk.

 

And you can do this one disk at a time because it doesn't matter if the data disks aren't the same size since each data disk is independent.

 

 

Merci pour le complément d'information

__________________________________________________________________________________________________________________________________________________________

 

Thank you for more information

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.