Details

    • Type: Sub-task Sub-task
    • Status: Resolved
    • Priority: Major Major
    • Resolution: Fixed
    • Fix Version/s: 1.2.0 beta 1
    • Component/s: Core
    • Labels:
      None

      Description

      The proposed changes are intended to preserve backwards compatibility:

      op behaviour deprecated warning?
      join Join the ring, use with -t to join at a specific token, or to add a token to an existing host
      ring prints the first token for each node, add -a to print all tokens
      move <new token> if the node only has 1 token then move it. Otherwise die with an error. deprecated
      removetoken status/force/<token> removes the node who owns <token> from the ring. use -t option to only remove the token from the node instead of the whole node.
      describering [keyspace] show all ranges and their endpoints
      getendpoints <keyspace> <cf> <key> Print the endpoints that own the key and also their list of tokens

      Edit0: Appended patch information.

      Patches

      Compare Raw diff Description
      01_admin_tools 01_admin_tools.patch Updated nodetool

      Note: These are branches managed with TopGit. If you are applying the patch output manually, you will either need to filter the TopGit metadata files (i.e. wget -O - <url> | filterdiff -x*.topdeps -x*.topmsg | patch -p1), or remove them afterward (rm .topmsg .topdeps).

        Activity

        Sam Overton created issue -
        Sam Overton made changes -
        Field Original Value New Value
        Component/s Core [ 12312978 ]
        Sam Overton made changes -
        Description The proposed changes are intended to preserve backwards compatibility:

        || op || behaviour || deprecated warning?
        | join | Join the ring, use with <tt>-t</tt> to join at a specific token, or to add a token to an existing host |
        | ring | prints the first token for each node, add <tt>-a</tt> to print all tokens |
        | move <new token> | if the node only has 1 token then move it. Otherwise die with an error. | *deprecated*
        | removetoken status/force/<token> | removes the node who owns <token> from the ring. use <tt>-t</tt> option to only remove the token from the node instead of the whole node. |
        | describering [keyspace] | show all ranges and their endpoints |
        | getendpoints <keyspace> <cf> <key> | Print the endpoints that own the key and also their list of tokens |
        The proposed changes are intended to preserve backwards compatibility:

        || op || behaviour || deprecated warning?
        | join | Join the ring, use with {{-t}} to join at a specific token, or to add a token to an existing host |
        | ring | prints the first token for each node, add {{-a}} to print all tokens |
        | move <new token> | if the node only has 1 token then move it. Otherwise die with an error. | *deprecated*
        | removetoken status/force/<token> | removes the node who owns <token> from the ring. use {{-t}} option to only remove the token from the node instead of the whole node. |
        | describering [keyspace] | show all ranges and their endpoints |
        | getendpoints <keyspace> <cf> <key> | Print the endpoints that own the key and also their list of tokens |
        Eric Evans made changes -
        Assignee Sam Overton [ soverton ] Eric Evans [ urandom ]
        Eric Evans made changes -
        Description The proposed changes are intended to preserve backwards compatibility:

        || op || behaviour || deprecated warning?
        | join | Join the ring, use with {{-t}} to join at a specific token, or to add a token to an existing host |
        | ring | prints the first token for each node, add {{-a}} to print all tokens |
        | move <new token> | if the node only has 1 token then move it. Otherwise die with an error. | *deprecated*
        | removetoken status/force/<token> | removes the node who owns <token> from the ring. use {{-t}} option to only remove the token from the node instead of the whole node. |
        | describering [keyspace] | show all ranges and their endpoints |
        | getendpoints <keyspace> <cf> <key> | Print the endpoints that own the key and also their list of tokens |
        The proposed changes are intended to preserve backwards compatibility:

        || op || behaviour || deprecated warning?
        | join | Join the ring, use with {{-t}} to join at a specific token, or to add a token to an existing host |
        | ring | prints the first token for each node, add {{-a}} to print all tokens |
        | move <new token> | if the node only has 1 token then move it. Otherwise die with an error. | *deprecated*
        | removetoken status/force/<token> | removes the node who owns <token> from the ring. use {{-t}} option to only remove the token from the node instead of the whole node. |
        | describering [keyspace] | show all ranges and their endpoints |
        | getendpoints <keyspace> <cf> <key> | Print the endpoints that own the key and also their list of tokens |

        _Edit0: Appended patch information._

        h3. Patches
        ||Compare||Raw diff||Description||
        |[01_admin_tools|https://github.com/acunu/cassandra/compare/top-bases/p/4125/01_admin_tools...p/4125/01_admin_tools]|[01_admin_tools.patch|https://github.com/acunu/cassandra/compare/top-bases/p/4125/01_admin_tools...p/4125/01_admin_tools.diff]|Updated nodetool|

        ----

        _Note: These are branches managed with TopGit. If you are applying the patch output manually, you will either need to filter the TopGit metadata files (i.e. {{wget -O - <url> | filterdiff -x*.topdeps -x*.topmsg | patch -p1}}), or remove them afterward ({{rm .topmsg .topdeps}})._
        Jonathan Ellis made changes -
        Reviewer brandon.williams
        Hide
        Brandon Williams added a comment -

        Can we also remove the deprecated methods here?

        Show
        Brandon Williams added a comment - Can we also remove the deprecated methods here?
        Hide
        Eric Evans added a comment -

        Can we also remove the deprecated methods here?

        Done in https://github.com/acunu/cassandra/commit/4486bdd

        Show
        Eric Evans added a comment - Can we also remove the deprecated methods here? Done in https://github.com/acunu/cassandra/commit/4486bdd
        Hide
        Brandon Williams added a comment -

        Ring still appears to print all the tokens without -a, not just the first, but otherwise looks good.

        Show
        Brandon Williams added a comment - Ring still appears to print all the tokens without -a, not just the first, but otherwise looks good.
        Hide
        Brandon Williams added a comment - - edited

        Also the 'owns' column in clusterinfo seems to have been broken (but I know it used to work) as it always reports zero now. On that note, perhaps we should think about combining ring and clusterinfo together, since depending on your setup and what you're doing you may have to go back and forth to get what you need.

        Show
        Brandon Williams added a comment - - edited Also the 'owns' column in clusterinfo seems to have been broken (but I know it used to work) as it always reports zero now. On that note, perhaps we should think about combining ring and clusterinfo together, since depending on your setup and what you're doing you may have to go back and forth to get what you need.
        Hide
        Eric Evans added a comment -

        Ring still appears to print all the tokens without -a, not just the first, but otherwise looks good.

        Did it support a -a argument at some point?

        Also the 'owns' column in clusterinfo seems to have been broken (but I know it used to work) as it always reports zero now.

        Right, probably more fall out from merging CASSANDRA-3047; Will have a look.

        ...perhaps we should think about combining ring and clusterinfo together, since depending on your setup and what you're doing you may have to go back and forth to get what you need.

        I'm of the opinion that it doesn't make sense to visualize the ring anymore. Perhaps it would be enough to incorporate some sort of per-DC ownership display into clusterinfo.

        Show
        Eric Evans added a comment - Ring still appears to print all the tokens without -a, not just the first, but otherwise looks good. Did it support a -a argument at some point? Also the 'owns' column in clusterinfo seems to have been broken (but I know it used to work) as it always reports zero now. Right, probably more fall out from merging CASSANDRA-3047 ; Will have a look. ...perhaps we should think about combining ring and clusterinfo together, since depending on your setup and what you're doing you may have to go back and forth to get what you need. I'm of the opinion that it doesn't make sense to visualize the ring anymore. Perhaps it would be enough to incorporate some sort of per-DC ownership display into clusterinfo.
        Hide
        Brandon Williams added a comment -

        Did it support a -a argument at some point?

        I don't think so, but it's in the description and might be a good idea, since people used to pre-vnode cassandra might be a bit shocked when thousands of lines scroll by when running it

        I'm of the opinion that it doesn't make sense to visualize the ring anymore. Perhaps it would be enough to incorporate some sort of per-DC ownership display into clusterinfo.

        I was thinking more of upgrades that aren't on vnodes yet. If you want to move a node, you need ring to get the token. If you want to remove a node, you need clusterinfo to get the hostid. And then if you want to know dc/rack info, you're back to ring (even with vnodes, except now you get scrollshocked.) It just seems like it'd be nice to get all the info you need from one place, the way ring currently works.

        Show
        Brandon Williams added a comment - Did it support a -a argument at some point? I don't think so, but it's in the description and might be a good idea, since people used to pre-vnode cassandra might be a bit shocked when thousands of lines scroll by when running it I'm of the opinion that it doesn't make sense to visualize the ring anymore. Perhaps it would be enough to incorporate some sort of per-DC ownership display into clusterinfo. I was thinking more of upgrades that aren't on vnodes yet. If you want to move a node, you need ring to get the token. If you want to remove a node, you need clusterinfo to get the hostid. And then if you want to know dc/rack info, you're back to ring (even with vnodes, except now you get scrollshocked.) It just seems like it'd be nice to get all the info you need from one place, the way ring currently works.
        Hide
        Eric Evans added a comment -
        Show
        Eric Evans added a comment - clusterinfo ownership fixed in https://github.com/acunu/cassandra/commit/bc9cae9
        Hide
        Eric Evans added a comment -

        [the -a arg is] in the description and might be a good idea, since people used to pre-vnode cassandra might be a bit shocked when thousands of lines scroll by when running it

        We could deprecate it, run it with a deprecation warning if num_tokens == 1, and force an additional argument to be passed (--yes-I-want-to-be-scroll-shocked) when num_tokens > 1.

        The problem I have with showing only one token is that it's very misleading. The "ring" has always been used to visualize topology in a world where there was a 1:1 relationship between node and token. Displaying our classic ring view with that 1:1 mapping when there are say hundreds of tokens seems almost certain to cause confusion.

        If you want to move a node, you need ring to get the token. If you want to remove a node, you need clusterinfo to get the hostid. And then if you want to know dc/rack info, you're back to ring (even with vnodes, except now you get scrollshocked.) It just seems like it'd be nice to get all the info you need from one place, the way ring currently works.

        What if we removed "Load" from clusterinfo to free up a little real estate (that seems like something people could get form info), added a "rack" column, and then page the list by DC the way CASSANDRA-3047 aimed to?

        Datacenter: DC1
        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
        Address          Rack   Status  State    Tokens  Owns    Host ID
        127.0.1.3        rack1  Up      Normal   8       16.00%   aaaa59d9-e5f9-445e-bb4b-a29a55075698
        127.0.2.2        rack2  Up      Normal   8       16.00%   bbbb31c5-e7eb-43c5-8393-2b71cd90d361
        127.0.3.1        rack3  Up      Normal   8       16.00%   cccc0bee-7292-4adb-9eb5-1f3bfd6d6d92
        
        Datacenter: DC2
        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
        Address          Rack   Status  State    Tokens  Owns    Host ID
        127.1.1.3        rack1  Up      Normal   8       16.00%   dddd59d9-e5f9-445e-bb4b-a29a55075698
        127.1.2.2        rack2  Up      Normal   8       16.00%   eeee31c5-e7eb-43c5-8393-2b71cd90d361
        127.1.3.1        rack3  Up      Normal   8       16.00%   ffff0bee-7292-4adb-9eb5-1f3bfd6d6d92
        
        Show
        Eric Evans added a comment - [the -a arg is] in the description and might be a good idea, since people used to pre-vnode cassandra might be a bit shocked when thousands of lines scroll by when running it We could deprecate it, run it with a deprecation warning if num_tokens == 1, and force an additional argument to be passed (--yes-I-want-to-be-scroll-shocked) when num_tokens > 1. The problem I have with showing only one token is that it's very misleading. The "ring" has always been used to visualize topology in a world where there was a 1:1 relationship between node and token. Displaying our classic ring view with that 1:1 mapping when there are say hundreds of tokens seems almost certain to cause confusion. If you want to move a node, you need ring to get the token. If you want to remove a node, you need clusterinfo to get the hostid. And then if you want to know dc/rack info, you're back to ring (even with vnodes, except now you get scrollshocked.) It just seems like it'd be nice to get all the info you need from one place, the way ring currently works. What if we removed "Load" from clusterinfo to free up a little real estate (that seems like something people could get form info ), added a "rack" column, and then page the list by DC the way CASSANDRA-3047 aimed to? Datacenter: DC1 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Address Rack Status State Tokens Owns Host ID 127.0.1.3 rack1 Up Normal 8 16.00% aaaa59d9-e5f9-445e-bb4b-a29a55075698 127.0.2.2 rack2 Up Normal 8 16.00% bbbb31c5-e7eb-43c5-8393-2b71cd90d361 127.0.3.1 rack3 Up Normal 8 16.00% cccc0bee-7292-4adb-9eb5-1f3bfd6d6d92 Datacenter: DC2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Address Rack Status State Tokens Owns Host ID 127.1.1.3 rack1 Up Normal 8 16.00% dddd59d9-e5f9-445e-bb4b-a29a55075698 127.1.2.2 rack2 Up Normal 8 16.00% eeee31c5-e7eb-43c5-8393-2b71cd90d361 127.1.3.1 rack3 Up Normal 8 16.00% ffff0bee-7292-4adb-9eb5-1f3bfd6d6d92
        Hide
        Brandon Williams added a comment -

        What if we removed "Load" from clusterinfo to free up a little real estate (that seems like something people could get form info),

        I don't think people will be very happy if we take that away, they're pretty used to being able to see that for all the machines in one shot.

        added a "rack" column, and then page the list by DC the way CASSANDRA-3047 aimed to?

        I do like paging by DC, that least saves some horizontal real estate by using vertical instead.

        Show
        Brandon Williams added a comment - What if we removed "Load" from clusterinfo to free up a little real estate (that seems like something people could get form info), I don't think people will be very happy if we take that away, they're pretty used to being able to see that for all the machines in one shot. added a "rack" column, and then page the list by DC the way CASSANDRA-3047 aimed to? I do like paging by DC, that least saves some horizontal real estate by using vertical instead.
        Hide
        Eric Evans added a comment -

        I don't think people will be very happy if we take [the Load column] away, they're pretty used to being able to see that for all the machines in one shot.

        Is this because it's one aspect of monitoring / maintaining balance between the nodes? Do you think this will be less relevant now?

        Show
        Eric Evans added a comment - I don't think people will be very happy if we take [the Load column] away, they're pretty used to being able to see that for all the machines in one shot. Is this because it's one aspect of monitoring / maintaining balance between the nodes? Do you think this will be less relevant now?
        Hide
        Brandon Williams added a comment -

        Is this because it's one aspect of monitoring / maintaining balance between the nodes?

        More as a quick check since, even with vnodes, you can create a large imbalance via fat rows.

        Do you think this will be less relevant now?

        To some degree, but not completely.

        I think new users want to be able to see how much space their cluster is using at a glance, and foresee a lot of complaints if we remove it at this point. Also, it's easier to say "send me your ring output" than "send me the ring output, then the output of info from each node" and space used is still pertinent information to have, outside of balance issues.

        Show
        Brandon Williams added a comment - Is this because it's one aspect of monitoring / maintaining balance between the nodes? More as a quick check since, even with vnodes, you can create a large imbalance via fat rows. Do you think this will be less relevant now? To some degree, but not completely. I think new users want to be able to see how much space their cluster is using at a glance, and foresee a lot of complaints if we remove it at this point. Also, it's easier to say "send me your ring output" than "send me the ring output, then the output of info from each node" and space used is still pertinent information to have, outside of balance issues.
        Hide
        Brandon Williams added a comment - - edited

        I think we may still have a problem with ownership calculation:

        Address          Status  State    Load       Tokens  Owns (effective) Host ID
        10.179.65.102    Up      Normal   197.82 MB  1       66.67%           e1c24ebd-e9c0-40eb-9419-e510d849c3e0
        10.179.111.137   Up      Normal   197.95 MB  256     99.87%           c9a17642-35dd-4eb5-9881-9631a5590a40
        10.179.64.227    Up      Normal   197.6 MB   1       33.46%           574f6677-d7a5-405a-9fb8-5c4d829f0a10
        

        .137 has 256 tokens, but they should be contiguous (upgraded)

        Show
        Brandon Williams added a comment - - edited I think we may still have a problem with ownership calculation: Address Status State Load Tokens Owns (effective) Host ID 10.179.65.102 Up Normal 197.82 MB 1 66.67% e1c24ebd-e9c0-40eb-9419-e510d849c3e0 10.179.111.137 Up Normal 197.95 MB 256 99.87% c9a17642-35dd-4eb5-9881-9631a5590a40 10.179.64.227 Up Normal 197.6 MB 1 33.46% 574f6677-d7a5-405a-9fb8-5c4d829f0a10 .137 has 256 tokens, but they should be contiguous (upgraded)
        Hide
        Eric Evans added a comment -

        I think new users want to be able to see how much space their cluster is using at a glance, and foresee a lot of complaints if we remove it at this point. Also, it's easier to say "send me your ring output" than "send me the ring output, then the output of info from each node" and space used is still pertinent information to have, outside of balance issues.

        I think we're on the same page wrt that point, all-on-one vs. spread-over-two, (s/ring output/clusterinfo output/ though). I just wanted to test the assertion that all of that information is still necessary.

        I'm sure it can be made compact enough to display everything.

        Show
        Eric Evans added a comment - I think new users want to be able to see how much space their cluster is using at a glance, and foresee a lot of complaints if we remove it at this point. Also, it's easier to say "send me your ring output" than "send me the ring output, then the output of info from each node" and space used is still pertinent information to have, outside of balance issues. I think we're on the same page wrt that point, all-on-one vs. spread-over-two, (s/ring output/clusterinfo output/ though). I just wanted to test the assertion that all of that information is still necessary. I'm sure it can be made compact enough to display everything.
        Hide
        Eric Evans added a comment -

        How about something like this?

        Status=Up/Down
        |/ State=Normal/Leaving/Joining/Moving
        ++  Address          Load      Tokens  Owns   Host ID                               Rack
        UN  222.222.222.221  999.9 KB  256     33.3%  2ed2bb57-5541-478e-be55-cf4a800fca87  rack1
        UL  222.222.222.222  1 MB      256     33.3%  528d3b78-305d-4319-bedd-393927fedb48  rack2
        UN  222.222.222.222  999 MB    256     33.3%  78843aba-6131-4d00-b38c-306086c5dcc7  rack3
        

        With more than one datacenter:

        Datacenter: Greece
        ==================
        Status=Up/Down
        |/ State=Normal/Leaving/Joining/Moving
        ++  Address          Load      Tokens  Owns   Host ID                               Rack
        UN  222.222.222.221  999.9 KB  256     11.1%  2ed2bb57-5541-478e-be55-cf4a800fca87  rack1
        UL  222.222.222.222  1 MB      256     11.1%  528d3b78-305d-4319-bedd-393927fedb48  rack2
        UN  222.222.222.222  999 MB    256     11.1%  78843aba-6131-4d00-b38c-306086c5dcc7  rack3
         
        Datacenter: CostaRica
        =====================
        Status=Up/Down
        |/ State=Normal/Leaving/Joining/Moving
        ++  Address          Load      Tokens  Owns   Host ID                               Rack
        UN  222.223.222.221  999.9 KB  256     11.1%  aed2bb57-5541-478e-be55-cf4a800fca87  rack11
        UL  222.223.222.222  1 MB      256     11.1%  b28d3b78-305d-4319-bedd-393927fedb48  rack22
        UN  222.223.222.222  999 MB    256     11.1%  c8843aba-6131-4d00-b38c-306086c5dcc7  rack33
         
        Datacenter: BritishColumbia
        ===========================
        Status=Up/Down
        |/ State=Normal/Leaving/Joining/Moving
        ++  Address          Load      Tokens  Owns   Host ID                               Rack
        UN  222.224.222.221  999.9 KB  256     11.1%  ded2bb57-5541-478e-be55-cf4a800fca87  rack12
        UL  222.224.222.222  1 MB      256     11.1%  e28d3b78-305d-4319-bedd-393927fedb48  rack22
        UN  222.224.222.222  999 MB    256     11.1%  f8843aba-6131-4d00-b38c-306086c5dcc7  rack33
        
        Show
        Eric Evans added a comment - How about something like this? Status=Up/Down |/ State=Normal/Leaving/Joining/Moving ++ Address Load Tokens Owns Host ID Rack UN 222.222.222.221 999.9 KB 256 33.3% 2ed2bb57-5541-478e-be55-cf4a800fca87 rack1 UL 222.222.222.222 1 MB 256 33.3% 528d3b78-305d-4319-bedd-393927fedb48 rack2 UN 222.222.222.222 999 MB 256 33.3% 78843aba-6131-4d00-b38c-306086c5dcc7 rack3 With more than one datacenter: Datacenter: Greece ================== Status=Up/Down |/ State=Normal/Leaving/Joining/Moving ++ Address Load Tokens Owns Host ID Rack UN 222.222.222.221 999.9 KB 256 11.1% 2ed2bb57-5541-478e-be55-cf4a800fca87 rack1 UL 222.222.222.222 1 MB 256 11.1% 528d3b78-305d-4319-bedd-393927fedb48 rack2 UN 222.222.222.222 999 MB 256 11.1% 78843aba-6131-4d00-b38c-306086c5dcc7 rack3 Datacenter: CostaRica ===================== Status=Up/Down |/ State=Normal/Leaving/Joining/Moving ++ Address Load Tokens Owns Host ID Rack UN 222.223.222.221 999.9 KB 256 11.1% aed2bb57-5541-478e-be55-cf4a800fca87 rack11 UL 222.223.222.222 1 MB 256 11.1% b28d3b78-305d-4319-bedd-393927fedb48 rack22 UN 222.223.222.222 999 MB 256 11.1% c8843aba-6131-4d00-b38c-306086c5dcc7 rack33 Datacenter: BritishColumbia =========================== Status=Up/Down |/ State=Normal/Leaving/Joining/Moving ++ Address Load Tokens Owns Host ID Rack UN 222.224.222.221 999.9 KB 256 11.1% ded2bb57-5541-478e-be55-cf4a800fca87 rack12 UL 222.224.222.222 1 MB 256 11.1% e28d3b78-305d-4319-bedd-393927fedb48 rack22 UN 222.224.222.222 999 MB 256 11.1% f8843aba-6131-4d00-b38c-306086c5dcc7 rack33
        Hide
        Brandon Williams added a comment -

        I like this, though maybe we could omit printing the status/state key all the time and put that in the help instead. It'd be nice to also print the first token, that way legacy machines with just one token achieve the 'one stop shop' goal.

        Show
        Brandon Williams added a comment - I like this, though maybe we could omit printing the status/state key all the time and put that in the help instead. It'd be nice to also print the first token, that way legacy machines with just one token achieve the 'one stop shop' goal.
        Hide
        Eric Evans added a comment - - edited

        Adding a token column that is conditionally printed when num_tokens == 1

         Datacenter: Greece
         ==================
         Status=Up/Down
         |/ State=Normal/Leaving/Joining/Moving
         ==  Address          Load      Owns   Host ID                               Token                                  Rack
         UN  222.222.222.221  999.9 KB  11.1%  2ed2bb57-5541-478e-be55-cf4a800fca87  0                                      rack1
         UL  222.222.222.222  1 MB      11.1%  528d3b78-305d-4319-bedd-393927fedb48  886151997189943915269204706853563050   rack2
         UN  222.222.222.222  999 MB    11.1%  78843aba-6131-4d00-b38c-306086c5dcc7  1772303994379887830538409413707126100  rack3
         
         Datacenter: CostaRica
         =====================
         Status=Up/Down
         |/ State=Normal/Leaving/Joining/Moving
         ==  Address          Load      Owns   Host ID                               Token                                   Rack
         UN  222.223.222.221  999.9 KB  11.1%  aed2bb57-5541-478e-be55-cf4a800fca87  9747671969089383067961251775389193550   rack11
         UL  222.223.222.222  1 MB      11.1%  b28d3b78-305d-4319-bedd-393927fedb48  10633823966279326983230456482242756600  rack22
         UN  222.223.222.222  999 MB    11.1%  c8843aba-6131-4d00-b38c-306086c5dcc7  11519975963469270898499661189096319650  rack33
         
         Datacenter: BritishColumbia
         ===========================
         Status=Up/Down
         |/ State=Normal/Leaving/Joining/Moving
         ==  Address          Load      Owns   Host ID                               Token                                   Rack
         UN  222.224.222.221  999.9 KB  11.1%  ded2bb57-5541-478e-be55-cf4a800fca87  12406127960659214813768865895949882700  rack12
         UL  222.224.222.222  1 MB      11.1%  e28d3b78-305d-4319-bedd-393927fedb48  13292279957849158729038070602803445750  rack22
         UN  222.224.222.222  999 MB    11.1%  f8843aba-6131-4d00-b38c-306086c5dcc7  14178431955039102644307275309657008800  rack33
        

        Not a huge fan of this myself.

        Show
        Eric Evans added a comment - - edited Adding a token column that is conditionally printed when num_tokens == 1 Datacenter: Greece ================== Status=Up/Down |/ State=Normal/Leaving/Joining/Moving == Address Load Owns Host ID Token Rack UN 222.222.222.221 999.9 KB 11.1% 2ed2bb57-5541-478e-be55-cf4a800fca87 0 rack1 UL 222.222.222.222 1 MB 11.1% 528d3b78-305d-4319-bedd-393927fedb48 886151997189943915269204706853563050 rack2 UN 222.222.222.222 999 MB 11.1% 78843aba-6131-4d00-b38c-306086c5dcc7 1772303994379887830538409413707126100 rack3 Datacenter: CostaRica ===================== Status=Up/Down |/ State=Normal/Leaving/Joining/Moving == Address Load Owns Host ID Token Rack UN 222.223.222.221 999.9 KB 11.1% aed2bb57-5541-478e-be55-cf4a800fca87 9747671969089383067961251775389193550 rack11 UL 222.223.222.222 1 MB 11.1% b28d3b78-305d-4319-bedd-393927fedb48 10633823966279326983230456482242756600 rack22 UN 222.223.222.222 999 MB 11.1% c8843aba-6131-4d00-b38c-306086c5dcc7 11519975963469270898499661189096319650 rack33 Datacenter: BritishColumbia =========================== Status=Up/Down |/ State=Normal/Leaving/Joining/Moving == Address Load Owns Host ID Token Rack UN 222.224.222.221 999.9 KB 11.1% ded2bb57-5541-478e-be55-cf4a800fca87 12406127960659214813768865895949882700 rack12 UL 222.224.222.222 1 MB 11.1% e28d3b78-305d-4319-bedd-393927fedb48 13292279957849158729038070602803445750 rack22 UN 222.224.222.222 999 MB 11.1% f8843aba-6131-4d00-b38c-306086c5dcc7 14178431955039102644307275309657008800 rack33 Not a huge fan of this myself.
        Hide
        Brandon Williams added a comment -

        lgtm.

        Show
        Brandon Williams added a comment - lgtm.
        Hide
        Eric Evans added a comment -

        OK, this output is implemented in: https://github.com/acunu/cassandra/commit/30710fb

        As of yet, I haven't figured out the strange ownership values (other than it being more fallout from CASSANDRA-3047).

        Show
        Eric Evans added a comment - OK, this output is implemented in: https://github.com/acunu/cassandra/commit/30710fb As of yet, I haven't figured out the strange ownership values (other than it being more fallout from CASSANDRA-3047 ).
        Hide
        Eric Evans added a comment -

        The ownership calculation should be fixed now.

        Show
        Eric Evans added a comment - The ownership calculation should be fixed now.
        Hide
        Brandon Williams added a comment -

        +1

        Show
        Brandon Williams added a comment - +1
        Eric Evans made changes -
        Fix Version/s 1.2 [ 12319262 ]
        Hide
        Eric Evans added a comment -

        committed to trunk; thanks everyone!

        Show
        Eric Evans added a comment - committed to trunk; thanks everyone!
        Eric Evans made changes -
        Status Open [ 1 ] Resolved [ 5 ]
        Resolution Fixed [ 1 ]
        Gavin made changes -
        Workflow no-reopen-closed, patch-avail [ 12661239 ] patch-available, re-open possible [ 12749543 ]
        Gavin made changes -
        Workflow patch-available, re-open possible [ 12749543 ] reopen-resolved, no closed status, patch-avail, testing [ 12757091 ]
        Transition Time In Source Status Execution Times Last Executer Last Execution Date
        Open Open Resolved Resolved
        104d 5h 2m 1 Eric Evans 18/Jul/12 20:16

          People

          • Assignee:
            Eric Evans
            Reporter:
            Sam Overton
            Reviewer:
            Brandon Williams
          • Votes:
            0 Vote for this issue
            Watchers:
            1 Start watching this issue

            Dates

            • Created:
              Updated:
              Resolved:

              Development