Conversation
This commit adds a variety of real disk metrics for the block devices that back Elasticsearch data paths. A collection of statistics are read from /proc/diskstats and are used to report the raw metrics for operations and read/write bytes. Relates elastic#15915
📝 WalkthroughWalkthroughThis PR introduces Linux-specific device I/O statistics collection to Elasticsearch's filesystem monitoring. It adds device metadata tracking via /proc/self/mountinfo, collects I/O metrics from /proc/diskstats, and exposes aggregated and per-device statistics through a new IoStats data structure with caching integration. Changes
Sequence DiagramsequenceDiagram
participant Client as Client / API
participant FSService as FsService
participant FSProbe as FsProbe
participant ESFileStore as ESFileStore
participant ProcFS as /proc (Linux)
participant Cache as Stats Cache
Client->>FSService: getStats()
FSService->>Cache: check cached value
alt Cache expired
FSService->>FSProbe: stats(previousFsInfo)
FSProbe->>ESFileStore: get device numbers
ESFileStore->>ProcFS: read /proc/self/mountinfo
ProcFS-->>ESFileStore: major/minor device numbers
FSProbe->>ProcFS: read /proc/diskstats
ProcFS-->>FSProbe: disk statistics
FSProbe->>FSProbe: compute IoStats (delta from previous)
FSProbe-->>FSService: new FsInfo with IoStats
FSService->>Cache: store FsInfo
end
Cache-->>FSService: FsInfo
FSService-->>Client: FsInfo with io_stats
Estimated code review effort🎯 4 (Complex) | ⏱️ ~75 minutes Poem
🚥 Pre-merge checks | ✅ 2 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing touches
Comment |
|
@CodeRabbit full review |
✅ Actions performedFull review triggered. |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 3
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
core/src/main/java/org/elasticsearch/env/ESFileStore.java (1)
48-78: Initialize device numbers to-1when mountinfo lookup fails or on non‑Linux.Right now
majorDeviceNumber/minorDeviceNumberremain0if the mount point isn’t found (or on non‑Linux), which can be interpreted as a real device. Defaulting to-1and breaking on the first match avoids false positives.🛠️ Suggested fix
if (Constants.LINUX) { try { spins = IOUtils.spins(PathUtils.get(getMountPointLinux(in))); } catch (Exception e) { spins = null; } + majorDeviceNumber = -1; + minorDeviceNumber = -1; try { - final List<String> lines = Files.readAllLines(PathUtils.get("/proc/self/mountinfo")); + final String mountPoint = getMountPointLinux(in); + final List<String> lines = Files.readAllLines(PathUtils.get("/proc/self/mountinfo")); for (final String line : lines) { final String[] fields = line.trim().split("\\s+"); - final String mountPoint = fields[4]; - if (mountPoint.equals(getMountPointLinux(in))) { + if (fields.length > 4 && mountPoint.equals(fields[4])) { final String[] deviceNumbers = fields[2].split(":"); majorDeviceNumber = Integer.parseInt(deviceNumbers[0]); minorDeviceNumber = Integer.parseInt(deviceNumbers[1]); + break; } } } catch (Exception e) { majorDeviceNumber = -1; minorDeviceNumber = -1; } } else { spins = null; + majorDeviceNumber = -1; + minorDeviceNumber = -1; }
🤖 Fix all issues with AI agents
In `@core/src/main/java/org/elasticsearch/monitor/fs/FsInfo.java`:
- Around line 389-405: In IoStats.toXContent (inside FsInfo) the
builder.startObject("total") call is not closed: after writing totalOperations,
totalReadOperations, totalWriteOperations, totalReadKilobytes and
totalWriteKilobytes you must call builder.endObject() to close the "total"
object; locate the block that iterates devicesStats and the code that starts the
"total" object (references: devicesStats, DeviceStats, totalOperations,
totalReadOperations, totalWriteOperations, totalReadKilobytes,
totalWriteKilobytes) and insert a builder.endObject() immediately after those
field() calls so the JSON structure is balanced.
In `@core/src/main/java/org/elasticsearch/monitor/fs/FsService.java`:
- Around line 67-80: The refresh() implementation in FsInfoCache is incorrectly
passing the fixed initialValue to stats(...) causing deltas to be computed
against the first sample; change the design so refresh() uses the current cached
FsInfo instead of initialValue: add a protected accessor in SingleObjectCache
(e.g., protected T getNoRefresh() or getCached()) that returns the current
cached value without triggering a refresh, then update FsInfoCache.refresh() to
call stats(probe, getNoRefresh(), logger) (or, alternatively, add a
previousField in FsInfoCache and update it on each successful refresh) and
ensure any added accessor maintains the existing concurrency/visibility
guarantees of SingleObjectCache.
In `@docs/reference/cluster/nodes-stats.asciidoc`:
- Around line 124-171: The docs incorrectly describe fs.io_stats fields as
"averages between probes" or "since starting Elasticsearch" but the
implementation reports deltas between consecutive snapshots; update the wording
for fs.io_stats.devices and its children (fs.io_stats.devices.device_name,
.operations, .read_operations, .write_operations, .read_kilobytes,
.write_kilobytes) and the aggregate fields (fs.io_stats.operations,
fs.io_stats.read_operations, fs.io_stats.write_operations,
fs.io_stats.read_kilobytes, fs.io_stats.write_kilobytes) to state these values
represent deltas over the probe interval (i.e., change since the previous
probe), and clarify that aggregate fields are computed from per-device deltas
rather than being cumulative since node start or averaged.
🧹 Nitpick comments (1)
core/src/main/java/org/elasticsearch/monitor/fs/FsProbe.java (1)
83-108: Add defensive bounds checking for /proc/diskstats parsing to skip malformed lines instead of failing entirely.The code accesses array indices 0–9 to parse
/proc/diskstatsfields (major, minor, device name, reads completed, sectors read, writes completed, sectors written). While the/proc/diskstatsformat defines at least 20 fields on modern Linux kernels, no explicit bounds check preventsArrayIndexOutOfBoundsExceptionif a line has fewer fields. Currently, such an exception propagates to the generic catch block, causing the entire method to returnnulland lose IO statistics for all devices.Skipping malformed lines gracefully is more robust:
♻️ Proposed defensive parsing
for (String line : lines) { - String fields[] = line.trim().split("\\s+"); + String[] fields = line.trim().split("\\s+"); + if (fields.length < 10) { + continue; // skip malformed lines + } final int majorDeviceNumber = Integer.parseInt(fields[0]);
📜 Review details
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (12)
core/src/main/java/org/elasticsearch/env/ESFileStore.javacore/src/main/java/org/elasticsearch/env/NodeEnvironment.javacore/src/main/java/org/elasticsearch/monitor/MonitorService.javacore/src/main/java/org/elasticsearch/monitor/fs/FsInfo.javacore/src/main/java/org/elasticsearch/monitor/fs/FsProbe.javacore/src/main/java/org/elasticsearch/monitor/fs/FsService.javacore/src/main/resources/org/elasticsearch/bootstrap/security.policycore/src/test/java/org/elasticsearch/cluster/DiskUsageTests.javacore/src/test/java/org/elasticsearch/monitor/fs/DeviceStatsTests.javacore/src/test/java/org/elasticsearch/monitor/fs/FsProbeTests.javadocs/reference/cluster/nodes-stats.asciidoctest/framework/src/main/java/org/elasticsearch/cluster/MockInternalClusterInfoService.java
🧰 Additional context used
🧬 Code graph analysis (3)
core/src/main/java/org/elasticsearch/monitor/fs/FsProbe.java (1)
core/src/main/java/org/elasticsearch/env/NodeEnvironment.java (2)
NodeEnvironment(78-941)NodePath(79-131)
core/src/test/java/org/elasticsearch/monitor/fs/FsProbeTests.java (1)
core/src/main/java/org/elasticsearch/env/NodeEnvironment.java (1)
NodeEnvironment(78-941)
core/src/main/java/org/elasticsearch/monitor/fs/FsService.java (3)
core/src/main/java/org/elasticsearch/common/logging/ESLogger.java (1)
ESLogger(30-206)core/src/main/java/org/elasticsearch/common/settings/Setting.java (1)
Setting(76-970)core/src/main/java/org/elasticsearch/common/unit/TimeValue.java (1)
TimeValue(37-345)
🔇 Additional comments (21)
core/src/main/resources/org/elasticsearch/bootstrap/security.policy (1)
127-129: LGTM: scoped /proc/diskstats read permission.This is the minimal permission needed for Linux I/O stats and is narrowly scoped to the exact proc path.
core/src/test/java/org/elasticsearch/cluster/DiskUsageTests.java (2)
203-208: LGTM: FsInfo constructor updated for IoStats parameter.Passing
nullhere is consistent with the new signature for tests that don't exercise IO stats.
244-249: LGTM: FsInfo constructor updated for IoStats parameter.This keeps test setup aligned with the new FsInfo API.
core/src/main/java/org/elasticsearch/env/ESFileStore.java (1)
250-256: LGTM: lucene device attributes exposed.This cleanly surfaces device metadata to consumers via the lucene attribute view.
core/src/main/java/org/elasticsearch/env/NodeEnvironment.java (1)
90-104: LGTM: NodePath captures device numbers from lucene attributes.This aligns NodePath metadata with the new ESFileStore attributes.
core/src/main/java/org/elasticsearch/monitor/fs/FsInfo.java (6)
191-323: LGTM: DeviceStats delta computation and serialization.The structure cleanly carries current/previous counters and exposes derived fields.
411-419: LGTM: FsInfo now carries IoStats.The new field and constructor wiring are straightforward.
426-439: Verify stream compatibility for the newioStatsfield.
FsInfois serialized over the wire; adding an unconditional optional field can break mixed‑version clusters unless it’s gated by stream version. Please confirm version checks are in place (or that BWC isn’t required for this branch).
471-473: LGTM: IoStats accessor added.
491-495: LGTM: io_stats included in fs XContent output.
500-506: LGTM: io_stats field key added.test/framework/src/main/java/org/elasticsearch/cluster/MockInternalClusterInfoService.java (1)
77-77: LGTM!The
nullIoStats parameter correctly aligns with the updatedFsInfoconstructor signature. For mock/test purposes, omitting IO stats is appropriate since this creates fake disk usage data.core/src/main/java/org/elasticsearch/monitor/fs/FsProbe.java (3)
22-38: LGTM!Imports are appropriate for the new Linux-specific IO stats functionality.
49-69: LGTM!The updated
stats()method correctly:
- Accepts a
previousFsInfo for delta calculations- Guards IO stats collection behind
Constants.LINUX- Filters out invalid device numbers (-1)
120-123: LGTM!The method is appropriately:
- Package-private for test overriding
- Annotated with
@SuppressForbiddenfor legitimate system file access- Simple and focused
core/src/test/java/org/elasticsearch/monitor/fs/DeviceStatsTests.java (1)
28-60: LGTM!The test correctly validates DeviceStats delta calculations:
- Operations sum (reads + writes)
- Kilobyte conversion (sectors / 2, since sectors are 512 bytes)
Consider adding edge case tests in the future for scenarios like counter wrap-around or missing previous stats, but this is adequate for initial coverage.
core/src/test/java/org/elasticsearch/monitor/fs/FsProbeTests.java (2)
48-69: LGTM!The test correctly validates:
- Linux path: IoStats presence, DeviceStats initialization with -1 for previous values
- Non-Linux path: IoStats is null
Good platform-conditional test coverage.
89-175: LGTM!Excellent test coverage that validates:
- Mock
/proc/diskstatsinjection via subclass override- Two-sample delta calculation (first sample with no previous, second with previous)
- Per-device stats accuracy
- Aggregate totals correctness
The test thoroughly exercises the
ioStats()method's parsing and calculation logic.core/src/main/java/org/elasticsearch/monitor/fs/FsService.java (2)
46-52: LGTM!Constructor correctly initializes the cache with an initial probe reading (passing
nullfor first sample).
58-65: LGTM!The static helper appropriately encapsulates error handling with debug-level logging, preventing probe failures from crashing the service.
core/src/main/java/org/elasticsearch/monitor/MonitorService.java (1)
81-82: LGTM!No functional changes in this file - only a trailing blank line formatting adjustment.
✏️ Tip: You can disable this entire section by setting review_details to false in your review settings.
| @Override | ||
| public XContentBuilder toXContent(XContentBuilder builder, Params params) throws IOException { | ||
| if (devicesStats.length > 0) { | ||
| builder.startArray("devices"); | ||
| for (DeviceStats deviceStats : devicesStats) { | ||
| builder.startObject(); | ||
| deviceStats.toXContent(builder, params); | ||
| builder.endObject(); | ||
| } | ||
| builder.endArray(); | ||
| builder.startObject("total"); | ||
| builder.field(OPERATIONS, totalOperations); | ||
| builder.field(READ_OPERATIONS, totalReadOperations); | ||
| builder.field(WRITE_OPERATIONS, totalWriteOperations); | ||
| builder.field(READ_KILOBYTES, totalReadKilobytes); | ||
| builder.field(WRITE_KILOBYTES, totalWriteKilobytes); | ||
| } |
There was a problem hiding this comment.
Close the total object in IoStats.toXContent.
builder.startObject("total") is never closed, which leaves the JSON structure unbalanced and causes the outer io_stats/fs objects to close incorrectly.
🛠️ Suggested fix
builder.startObject("total");
builder.field(OPERATIONS, totalOperations);
builder.field(READ_OPERATIONS, totalReadOperations);
builder.field(WRITE_OPERATIONS, totalWriteOperations);
builder.field(READ_KILOBYTES, totalReadKilobytes);
builder.field(WRITE_KILOBYTES, totalWriteKilobytes);
+ builder.endObject();
}
return builder;📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| @Override | |
| public XContentBuilder toXContent(XContentBuilder builder, Params params) throws IOException { | |
| if (devicesStats.length > 0) { | |
| builder.startArray("devices"); | |
| for (DeviceStats deviceStats : devicesStats) { | |
| builder.startObject(); | |
| deviceStats.toXContent(builder, params); | |
| builder.endObject(); | |
| } | |
| builder.endArray(); | |
| builder.startObject("total"); | |
| builder.field(OPERATIONS, totalOperations); | |
| builder.field(READ_OPERATIONS, totalReadOperations); | |
| builder.field(WRITE_OPERATIONS, totalWriteOperations); | |
| builder.field(READ_KILOBYTES, totalReadKilobytes); | |
| builder.field(WRITE_KILOBYTES, totalWriteKilobytes); | |
| } | |
| `@Override` | |
| public XContentBuilder toXContent(XContentBuilder builder, Params params) throws IOException { | |
| if (devicesStats.length > 0) { | |
| builder.startArray("devices"); | |
| for (DeviceStats deviceStats : devicesStats) { | |
| builder.startObject(); | |
| deviceStats.toXContent(builder, params); | |
| builder.endObject(); | |
| } | |
| builder.endArray(); | |
| builder.startObject("total"); | |
| builder.field(OPERATIONS, totalOperations); | |
| builder.field(READ_OPERATIONS, totalReadOperations); | |
| builder.field(WRITE_OPERATIONS, totalWriteOperations); | |
| builder.field(READ_KILOBYTES, totalReadKilobytes); | |
| builder.field(WRITE_KILOBYTES, totalWriteKilobytes); | |
| builder.endObject(); | |
| } |
🤖 Prompt for AI Agents
In `@core/src/main/java/org/elasticsearch/monitor/fs/FsInfo.java` around lines 389
- 405, In IoStats.toXContent (inside FsInfo) the builder.startObject("total")
call is not closed: after writing totalOperations, totalReadOperations,
totalWriteOperations, totalReadKilobytes and totalWriteKilobytes you must call
builder.endObject() to close the "total" object; locate the block that iterates
devicesStats and the code that starts the "total" object (references:
devicesStats, DeviceStats, totalOperations, totalReadOperations,
totalWriteOperations, totalReadKilobytes, totalWriteKilobytes) and insert a
builder.endObject() immediately after those field() calls so the JSON structure
is balanced.
| private class FsInfoCache extends SingleObjectCache<FsInfo> { | ||
| public FsInfoCache(TimeValue interval, FsInfo initValue) { | ||
| super(interval, initValue); | ||
|
|
||
| private final FsInfo initialValue; | ||
|
|
||
| public FsInfoCache(TimeValue interval, FsInfo initialValue) { | ||
| super(interval, initialValue); | ||
| this.initialValue = initialValue; | ||
| } | ||
|
|
||
| @Override | ||
| protected FsInfo refresh() { | ||
| try { | ||
| return probe.stats(); | ||
| } catch (IOException ex) { | ||
| logger.warn("Failed to fetch fs stats - returning empty instance"); | ||
| return new FsInfo(0, null); | ||
| } | ||
| return stats(probe, initialValue, logger); | ||
| } | ||
|
|
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
# Find SingleObjectCache class definition
fd -t f "SingleObjectCache" --exec cat {} \;Repository: Signal65/elasticsearch-CodeRabbit
Length of output: 6270
🏁 Script executed:
# If not found, search for it by content
rg -t java "class SingleObjectCache" -A 20Repository: Signal65/elasticsearch-CodeRabbit
Length of output: 4691
🏁 Script executed:
# Also look for getNoRefresh method
rg -t java "getNoRefresh"Repository: Signal65/elasticsearch-CodeRabbit
Length of output: 59
🏁 Script executed:
# Search for the stats method to understand what the previous parameter does
rg -t java "private.*FsInfo.*stats\(" -A 10 core/src/main/java/org/elasticsearch/monitor/fs/FsService.javaRepository: Signal65/elasticsearch-CodeRabbit
Length of output: 443
🏁 Script executed:
# Find FsProbe class and its stats method
rg -t java "class FsProbe" -A 50 | head -100Repository: Signal65/elasticsearch-CodeRabbit
Length of output: 11112
🏁 Script executed:
# Search for probe.stats method implementation
rg -t java "public.*stats\(" core/src/main/java/org/elasticsearch/monitor/fs/ -A 10Repository: Signal65/elasticsearch-CodeRabbit
Length of output: 2317
🏁 Script executed:
# Look for FsInfo class and how deltas are calculated
rg -t java "class FsInfo" -A 30 | head -80Repository: Signal65/elasticsearch-CodeRabbit
Length of output: 4429
Bug: refresh() always uses initialValue instead of the current cached value for delta calculations.
The refresh() method passes initialValue (the first sample) as the previous parameter on every refresh, causing IO stats deltas to be computed against the initial sample rather than the previous sample. This breaks continuous delta tracking.
To fix this, FsInfoCache needs access to the currently cached FsInfo. However, SingleObjectCache does not expose a method to retrieve the cached value without triggering a refresh, and the cached field is private. The fix requires either:
- Adding a protected method to
SingleObjectCacheto access the cached value (e.g.,protected T getNoRefresh()) - Storing the previous
FsInfoinFsInfoCacheitself and updating it during refresh
🤖 Prompt for AI Agents
In `@core/src/main/java/org/elasticsearch/monitor/fs/FsService.java` around lines
67 - 80, The refresh() implementation in FsInfoCache is incorrectly passing the
fixed initialValue to stats(...) causing deltas to be computed against the first
sample; change the design so refresh() uses the current cached FsInfo instead of
initialValue: add a protected accessor in SingleObjectCache (e.g., protected T
getNoRefresh() or getCached()) that returns the current cached value without
triggering a refresh, then update FsInfoCache.refresh() to call stats(probe,
getNoRefresh(), logger) (or, alternatively, add a previousField in FsInfoCache
and update it on each successful refresh) and ensure any added accessor
maintains the existing concurrency/visibility guarantees of SingleObjectCache.
| `fs.io_stats.devices` (Linux only):: | ||
| Array of disk metrics for each device that is backing an | ||
| Elasticsearch data path. These disk metrics are probed periodically | ||
| and averages between the last probe and the current probe are | ||
| computed. | ||
|
|
||
| `fs.io_stats.devices.device_name` (Linux only):: | ||
| The Linux device name. | ||
|
|
||
| `fs.io_stats.devices.operations` (Linux only):: | ||
| The total number of read and write operations for the device | ||
| completed since starting Elasticsearch. | ||
|
|
||
| `fs.io_stats.devices.read_operations` (Linux only):: | ||
| The total number of read operations for the device completed since | ||
| starting Elasticsearch. | ||
|
|
||
| `fs.io_stats.devices.write_operations` (Linux only):: | ||
| The total number of write operations for the device completed since | ||
| starting Elasticsearch. | ||
|
|
||
| `fs.io_stats.devices.read_kilobytes` (Linux only):: | ||
| The total number of kilobytes read for the device since starting | ||
| Elasticsearch. | ||
|
|
||
| `fs.io_stats.devices.write_kilobytes` (Linux only):: | ||
| The total number of kilobytes written for the device since | ||
| starting Elasticsearch. | ||
|
|
||
| `fs.io_stats.operations` (Linux only):: | ||
| The total number of read and write operations across all devices | ||
| used by Elasticsearch completed since starting Elasticsearch. | ||
|
|
||
| `fs.io_stats.read_operations` (Linux only):: | ||
| The total number of read operations for across all devices used by | ||
| Elasticsearch completed since starting Elasticsearch. | ||
|
|
||
| `fs.io_stats.write_operations` (Linux only):: | ||
| The total number of write operations across all devices used by | ||
| Elasticsearch completed since starting Elasticsearch. | ||
|
|
||
| `fs.io_stats.read_kilobytes` (Linux only):: | ||
| The total number of kilobytes read across all devices used by | ||
| Elasticsearch since starting Elasticsearch. | ||
|
|
||
| `fs.io_stats.write_kilobytes` (Linux only):: | ||
| The total number of kilobytes written across all devices used by | ||
| Elasticsearch since starting Elasticsearch. |
There was a problem hiding this comment.
Clarify io_stats semantics (delta vs cumulative/average).
The text currently mentions “averages between probes” and “since starting Elasticsearch,” but the implementation computes deltas between snapshots. Updating the wording avoids misleading users.
✏️ Suggested doc edits
- Elasticsearch data path. These disk metrics are probed periodically
- and averages between the last probe and the current probe are
- computed.
+ Elasticsearch data path. These disk metrics are probed periodically
+ and deltas between the last probe and the current probe are
+ reported.
- The total number of read and write operations for the device
- completed since starting Elasticsearch.
+ The total number of read and write operations for the device
+ completed since the previous probe.
- The total number of read operations for the device completed since
- starting Elasticsearch.
+ The total number of read operations for the device completed since
+ the previous probe.
- The total number of write operations for the device completed since
- starting Elasticsearch.
+ The total number of write operations for the device completed since
+ the previous probe.
- The total number of kilobytes read for the device since starting
- Elasticsearch.
+ The total number of kilobytes read for the device since the
+ previous probe.
- The total number of kilobytes written for the device since
- starting Elasticsearch.
+ The total number of kilobytes written for the device since the
+ previous probe.
- The total number of read and write operations across all devices
- used by Elasticsearch completed since starting Elasticsearch.
+ The total number of read and write operations across all devices
+ used by Elasticsearch completed since the previous probe.
- The total number of read operations for across all devices used by
- Elasticsearch completed since starting Elasticsearch.
+ The total number of read operations for across all devices used by
+ Elasticsearch completed since the previous probe.
- The total number of write operations across all devices used by
- Elasticsearch completed since starting Elasticsearch.
+ The total number of write operations across all devices used by
+ Elasticsearch completed since the previous probe.
- The total number of kilobytes read across all devices used by
- Elasticsearch since starting Elasticsearch.
+ The total number of kilobytes read across all devices used by
+ Elasticsearch since the previous probe.
- The total number of kilobytes written across all devices used by
- Elasticsearch since starting Elasticsearch.
+ The total number of kilobytes written across all devices used by
+ Elasticsearch since the previous probe.📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| `fs.io_stats.devices` (Linux only):: | |
| Array of disk metrics for each device that is backing an | |
| Elasticsearch data path. These disk metrics are probed periodically | |
| and averages between the last probe and the current probe are | |
| computed. | |
| `fs.io_stats.devices.device_name` (Linux only):: | |
| The Linux device name. | |
| `fs.io_stats.devices.operations` (Linux only):: | |
| The total number of read and write operations for the device | |
| completed since starting Elasticsearch. | |
| `fs.io_stats.devices.read_operations` (Linux only):: | |
| The total number of read operations for the device completed since | |
| starting Elasticsearch. | |
| `fs.io_stats.devices.write_operations` (Linux only):: | |
| The total number of write operations for the device completed since | |
| starting Elasticsearch. | |
| `fs.io_stats.devices.read_kilobytes` (Linux only):: | |
| The total number of kilobytes read for the device since starting | |
| Elasticsearch. | |
| `fs.io_stats.devices.write_kilobytes` (Linux only):: | |
| The total number of kilobytes written for the device since | |
| starting Elasticsearch. | |
| `fs.io_stats.operations` (Linux only):: | |
| The total number of read and write operations across all devices | |
| used by Elasticsearch completed since starting Elasticsearch. | |
| `fs.io_stats.read_operations` (Linux only):: | |
| The total number of read operations for across all devices used by | |
| Elasticsearch completed since starting Elasticsearch. | |
| `fs.io_stats.write_operations` (Linux only):: | |
| The total number of write operations across all devices used by | |
| Elasticsearch completed since starting Elasticsearch. | |
| `fs.io_stats.read_kilobytes` (Linux only):: | |
| The total number of kilobytes read across all devices used by | |
| Elasticsearch since starting Elasticsearch. | |
| `fs.io_stats.write_kilobytes` (Linux only):: | |
| The total number of kilobytes written across all devices used by | |
| Elasticsearch since starting Elasticsearch. | |
| `fs.io_stats.devices` (Linux only):: | |
| Array of disk metrics for each device that is backing an | |
| Elasticsearch data path. These disk metrics are probed periodically | |
| and deltas between the last probe and the current probe are | |
| reported. | |
| `fs.io_stats.devices.device_name` (Linux only):: | |
| The Linux device name. | |
| `fs.io_stats.devices.operations` (Linux only):: | |
| The total number of read and write operations for the device | |
| completed since the previous probe. | |
| `fs.io_stats.devices.read_operations` (Linux only):: | |
| The total number of read operations for the device completed since | |
| the previous probe. | |
| `fs.io_stats.devices.write_operations` (Linux only):: | |
| The total number of write operations for the device completed since | |
| the previous probe. | |
| `fs.io_stats.devices.read_kilobytes` (Linux only):: | |
| The total number of kilobytes read for the device since the | |
| previous probe. | |
| `fs.io_stats.devices.write_kilobytes` (Linux only):: | |
| The total number of kilobytes written for the device since the | |
| previous probe. | |
| `fs.io_stats.operations` (Linux only):: | |
| The total number of read and write operations across all devices | |
| used by Elasticsearch completed since the previous probe. | |
| `fs.io_stats.read_operations` (Linux only):: | |
| The total number of read operations for across all devices used by | |
| Elasticsearch completed since the previous probe. | |
| `fs.io_stats.write_operations` (Linux only):: | |
| The total number of write operations across all devices used by | |
| Elasticsearch completed since the previous probe. | |
| `fs.io_stats.read_kilobytes` (Linux only):: | |
| The total number of kilobytes read across all devices used by | |
| Elasticsearch since the previous probe. | |
| `fs.io_stats.write_kilobytes` (Linux only):: | |
| The total number of kilobytes written across all devices used by | |
| Elasticsearch since the previous probe. |
🤖 Prompt for AI Agents
In `@docs/reference/cluster/nodes-stats.asciidoc` around lines 124 - 171, The docs
incorrectly describe fs.io_stats fields as "averages between probes" or "since
starting Elasticsearch" but the implementation reports deltas between
consecutive snapshots; update the wording for fs.io_stats.devices and its
children (fs.io_stats.devices.device_name, .operations, .read_operations,
.write_operations, .read_kilobytes, .write_kilobytes) and the aggregate fields
(fs.io_stats.operations, fs.io_stats.read_operations,
fs.io_stats.write_operations, fs.io_stats.read_kilobytes,
fs.io_stats.write_kilobytes) to state these values represent deltas over the
probe interval (i.e., change since the previous probe), and clarify that
aggregate fields are computed from per-device deltas rather than being
cumulative since node start or averaged.
2 participants
PR_015
Summary by CodeRabbit
New Features
Documentation
✏️ Tip: You can customize this high-level summary in your review settings.