From 646aa0284e833903ab23ca1edb6677dffdefca94 Mon Sep 17 00:00:00 2001
From: Iustin Pop <iustin@google.com>
Date: Tue, 30 Nov 2010 15:42:14 +0000
Subject: [PATCH] hbal: implement handling of multi-group clusters
On a single-group cluster, we proceed as before. On multi-group
clusters, we require selection of the desired group (currently via UUID
only).
Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Balazs Lecz <leczb@google.com>
---
hbal.hs | 44 ++++++++++++++++++++++++++++++++++++++------
man/hbal.rst | 5 +++++
2 files changed, 43 insertions(+), 6 deletions(-)
diff --git a/hbal.hs b/hbal.hs
index 59edde484..f8ad8d065 100644
--- a/hbal.hs
+++ b/hbal.hs
@@ -28,7 +28,7 @@ module Main (main) where
import Control.Concurrent (threadDelay)
import Control.Exception (bracket)
import Data.List
-import Data.Maybe (isJust, fromJust)
+import Data.Maybe (isJust, isNothing, fromJust)
import Data.IORef
import Monad
import System (exitWith, ExitCode(..))
@@ -66,6 +66,7 @@ options =
, oRapiMaster
, oLuxiSocket
, oExecJobs
+ , oGroup
, oMaxSolLength
, oVerbose
, oQuiet
@@ -225,7 +226,7 @@ main = do
verbose = optVerbose opts
shownodes = optShowNodes opts
- (fixed_nl, il, ctags) <- loadExternalData opts
+ (fixed_nl, ilf, ctags) <- loadExternalData opts
let offline_names = optOffline opts
all_nodes = Container.elems fixed_nl
@@ -238,7 +239,7 @@ main = do
all_nodes
m_cpu = optMcpu opts
m_dsk = optMdsk opts
- csf = commonSuffix fixed_nl il
+ csf = commonSuffix fixed_nl ilf
when (length offline_wrong > 0) $ do
hPrintf stderr "Wrong node name(s) set as offline: %s\n"
@@ -248,21 +249,52 @@ main = do
let nm = Container.map (\n -> if Node.idx n `elem` offline_indices
then Node.setOffline n True
else n) fixed_nl
- nl = Container.map (flip Node.setMdsk m_dsk . flip Node.setMcpu m_cpu)
- nm
+ nlf = Container.map (flip Node.setMdsk m_dsk . flip Node.setMcpu m_cpu)
+ nm
when (not oneline && verbose > 1) $
putStrLn $ "Loaded cluster tags: " ++ intercalate "," ctags
- when (Container.size il == 0) $ do
+ when (Container.size ilf == 0) $ do
(if oneline then putStrLn $ formatOneline 0 0 0
else printf "Cluster is empty, exiting.\n")
exitWith ExitSuccess
+ let split_insts = Cluster.findSplitInstances nlf ilf
+ when (not . null $ split_insts) $ do
+ hPutStrLn stderr "Found instances belonging to multiple node groups:"
+ mapM_ (\i -> hPutStrLn stderr $ " " ++ Instance.name i) split_insts
+ hPutStrLn stderr "Aborting."
+ exitWith $ ExitFailure 1
+
+ let ngroups = Cluster.splitCluster nlf ilf
+ when (length ngroups > 1 && isNothing (optGroup opts)) $ do
+ hPutStrLn stderr "Found multiple node groups:"
+ mapM_ (hPutStrLn stderr . (" " ++) . fst ) ngroups
+ hPutStrLn stderr "Aborting."
+ exitWith $ ExitFailure 1
+
unless oneline $ printf "Loaded %d nodes, %d instances\n"
+ (Container.size nlf)
+ (Container.size ilf)
+
+ (guuid, (nl, il)) <- case optGroup opts of
+ Nothing -> return $ head ngroups
+ Just g -> case lookup g ngroups of
+ Nothing -> do
+ hPutStrLn stderr $ "Node group " ++ g ++
+ " not found. Node group list is:"
+ mapM_ (hPutStrLn stderr . (" " ++) . fst ) ngroups
+ hPutStrLn stderr "Aborting."
+ exitWith $ ExitFailure 1
+ Just cdata -> return (g, cdata)
+
+ unless oneline $ printf "Group size %d nodes, %d instances\n"
(Container.size nl)
(Container.size il)
+ putStrLn $ "Selected node group: " ++ guuid
+
when (length csf > 0 && not oneline && verbose > 1) $
printf "Note: Stripping common suffix of '%s' from names\n" csf
diff --git a/man/hbal.rst b/man/hbal.rst
index 8fe081cb6..7988aa3a1 100644
--- a/man/hbal.rst
+++ b/man/hbal.rst
@@ -436,6 +436,11 @@ The options that can be passed to the program are as follows:
number. For example, specifying *disk-ratio* as **0.25** means that
at least one quarter of disk space should be left free on nodes.
+-G *uuid*, --group=*uuid*
+ On an multi-group cluster, select this group for
+ processing. Otherwise hbal will abort, since it cannot balance
+ multiple groups at the same time.
+
-v, --verbose
Increase the output verbosity. Each usage of this option will
increase the verbosity (currently more than 2 doesn't make sense)
--
GitLab